- Email: [email protected]
Flavor Preferences Conditioned by High-Fat versus High-Carbohydrate Diets Vary as a Function of Session Length
Physiology & Behavior, Vol. 66, No. 3, pp. 389–395, 1999 © 1999 Elsevier Science Inc. Printed in the USA. All rights reserved 0031-9384/99/$–see front matter
PII S0031-9384(98)00274-1
Flavor Preferences Conditioned by High-Fat versus High-Carbohydrate Diets Vary as a Function of Session Length FRANÇOIS LUCAS AND ANTHONY SCLAFANI1 Department of Psychology, Brooklyn College and the Graduate School, The City University of New York, 2900 Bedford Ave, Brooklyn, NY 11210-2889 Received 3 August 1998; Accepted 1 September 1998 LUCAS, F. AND A. SCLAFANI. Flavor preferences conditioned by high-fat versus high-carbohydrate diets vary as a function of session length. PHYSIOL BEHAV 66(3) 389–395, 1999.—Intragastric (i.g.) infusions of fat and carbohydrate condition flavor preferences in rats, but different results have been obtained in studies using pure and mixed nutrient infusions. This experiment compared the preference conditioning effects of mixed high-carbohydrate (HC) and high-fat (HF) diet infusions during short-term and long-term sessions. In Experiment 1 food-deprived rats were given one flavored saccharin solution (CS1HC) paired with i.g. infusions of an HC liquid diet, a second flavor (CS1HF) paired with HF diet infusions, and a third flavor (CS2) paired with i.g. water infusions during 30-min one-bottle training sessions. In subsequent two-bottle tests (30 min/day), the rats preferred both CS1s to the CS2 and preferred the CS1HC to the CS1HF. In Experiment 2, the same rats were trained and tested with the CSs and paired infusions during 22 h/day sessions with chow available ad lib. Both CS1s were again preferred to the CS2, but now the CS1HF was preferred to the CS1HC. When given additional 30-min choice sessions in Experiment 3 the rats showed no reliable preference for the CS1HC versus CS1HF under food-deprived or ad lib conditions. In Experiment 4, the rats were given 22-h CS1HC versus CS1HF choice sessions every other day. They showed no reliable CS preference during the first 30 min of each session, but reliably preferred the CS1HF during the remaining 21.5 h. These findings indicate that previously reported differences in preferences conditioned by pure versus mixed nutrient infusions are due to training procedures (session length, deprivation state) rather than to the type of nutrient infusions per se. The rats displayed different CS1HF versus CS1HC preferences as a function of test duration even after being given both shortand long-term training. Thus, short-term choice tests do not always predict the long-term intakes and preferences for high-fat and high-carbohydrate foods. © 1999 Elsevier Science Inc. Flavor preferences
Gastric infusions
Long-term test
Short-term test
In another recent study (4) rats were trained using complete liquid diets that differed in their fat and carbohydrate content. In this case, the rats acquired a preference for a flavor (CS1HF) paired with i.g. infusions of a high-fat (HF, 60% fat kcal) diet over another flavor (CS1HC) paired with infusions of a high-carbohydrate (HC, 79% CHO kcal) diet. It would appear from these data that fat is less reinforcing than carbohydrate when presented in a pure form, but more reinforcing when presented within a mixed diet. However, in addition to the type of nutrient infusions (pure versus mixed), the two studies differed in other respects, which may have contributed to the opposing results. In particular, the rats infused with pure nutrients were food deprived and trained in short (30-min) sessions, while the rats infused with mixed diets were nondeprived and trained in long (22-h) sessions. The
IT is now well established that food preferences in rats are conditioned by the postingestive reinforcing effects of nutrients (2,12). In recent studies (4,8,9) we have compared the reinforcing actions of fat (F) and carbohydrate (C) by pairing the intake of different flavored solutions (conditioned stimuli, CS) with intragastric (i.g.) infusions that differed in their fat/ carbohydrate content. As reported in the previous article (8), rats trained with one flavor (CS1F) paired with i.g. fat (7.1% corn oil) and another flavor (CS1C) paired with i.g. carbohydrate (16% maltodextrin) subsequently displayed a 70% preference for the CS1C over the CS1F [see also (9)]. Both flavors, however, were preferred to a third flavor (CS2), which was paired with i.g. water infusions. These findings confirm and extend earlier data indicating that fat infusions are less reinforcing than isocaloric carbohydrate infusions (3,6,7,10). 1To
Food deprivation
whom requests for reprints should be addressed. E-mail: [email protected]
389
390
LUCAS AND SCLAFANI
present experiment was conducted to determine whether deprivation state or session length influences the preference conditioning effects of high-fat and high-carbohydrate mixed diets.
EXPERIMENT 1
Methods Subjects. Eleven adult female Sprague–Dawley rats bred in our laboratory from Charles River Laboratories (Wilmington, MA) stock were used. The animals weighed 210–310 g at the beginning of the experiment. The animals were individually housed in wire-mesh cages in a room maintained at 218C under a 12:12-h light:dark cycle (lights on at 0800 h). Surgery. The rats were surgically implanted with stainless steel gastric cannulas as described by Elizalde and Sclafani (3). The animals were allowed to recover from surgery for at least 1 week prior to the start of training. When not in use, the cannulas were kept closed with stainless steel screws. Apparatus. Intragastric infusions were accomplished using an “electronic esophagus” apparatus described in detail elsewhere (3). In brief, the rats were housed in standard stainless steel cages modified so that powdered chow was available from a food cup accessible through a hole in the back wall of the cage. Drinking fluids were available through stainless steel sipper tubes available through two 19-mm holes in the front of the cage. Trays below the sipper tubes collected spillage. A slot in the cage floor permitted a stainless steel spring containing two catheters to connect the rat’s gastric cannula to outputs of a dual-channel infusion swivel located below the cage. Plastic tubing connected the swivel’s inputs to two peristaltic infusion pumps. The infusion pumps were operated automatically by drinkometer circuits and a microcomputer whenever the rat licked one or the other of two sipper tubes; the computer stored the number of licks/min for offline drinking pattern analysis. The rate of infusion was 1.5 mL/min, and about 1 mL of fluid was infused i.g. for each 1 mL of fluid orally consumed. Test solutions and diets. The CSs consisted of 0.2% sodium saccharin (Sigma, St. Louis, MO) solutions flavored with 0.05% cherry, grape or orange unsweetened Kool-Aid (General Foods, White Plains, NY). Unflavored 0.2% saccharin and 2% sucrose10.2% saccharin solutions were also used for training purposes. The i.g. infusions contained water, a high-fat (HF) diet, or a high-carbohydrate (HC) diet. The HF and HC diets were prepared by adding extra carbohydrate (maltodextrin ) and/or fat emulsion (corn oil) to evaporated milk, and they contained similar amounts of protein and energy (2.1 kcal/mL; see Table 1). The flavors paired with water, HC diet, and HF diet are referred to as CS2, CS1HC, and CS1HF, respectively. The specific flavor paired with each infusion was counterbalanced across subjects. Intakes of CS solutions and i.g. diets were recorded in g (60.1 g) and converted to milliliters. The rats were fed a maintenance diet of powdered Purina Chow (No. 5001). Procedure. Two weeks after surgery, the rats were familiarized with the saccharin solution by giving them two-bottle access to water and the saccharin1sucrose solution for 2 days followed by two-bottle access to water and saccharin for 2 days. They were then transferred to the test cages and attached to the infusion apparatus. The rats were maintained on a food restriction schedule that provided 40 kcal/day of chow. Total daily energy (chow 1 i.g. diet) intake averaged approximately 55 kcal/day throughout the experiment. This represented about 80% of their ad lib chow intake and was compa-
TABLE 1 COMPOSITION OF HIGH-FAT (HF) AND HIGH-CARBOHYDRATE (HC) DIETS
Evaporated milk,* g Maltodextrin,† g Corn oil,‡ g Sodium stearoyl lactylate,§ g Water, g Total, g Volume, ml kcal/ml % protein kcal % carbohydrate kcal % fat kcal
HF
HC
497.2 124.7 98.9 1.3 277.9 1000 855 2.1 6.4 34.0 59.6
458.2 320.1 — — 221.6 1000 934 2.1 6.4 79.0 14.6
*Pathmark Supermarket Brand. †Maltrin M580, Grain Processing Corp., Muscatine, IA. ‡Mazola, Best Foods, Englewood Cliffs, NJ. §Emplex, American Products, Kansas City, MO.
rable to the energy intakes allowed in our previous study of pure fat and carbohydrate conditioning (8). The rats were trained to drink the unflavored saccharin solution during seven daily 30-min sessions in the early light period (~2 h after light onset). During the last three of these pretraining sessions, intake of the saccharin solution was paired with i.g. water infusions. This was followed by one-bottle flavor conditioning sessions. Half of the rats were given the CS1HC paired with i.g. HC diet infusions on Day 1, the CS2 paired with i.g. water on Day 2, and the CS1HF paired with i.g. HF diet on Day 3. The remaining rats were given the CS1HF on Day 1 and the CS1HC on Day 3. This 3-day pattern was repeated three times for a total of 12 training sessions. The first three training sessions were 2 h in length to ensure that all rats would sample the CS solutions; oral and i.g. intakes were limited to a maximum of 10 mL each. The remaining training sessions were 30 min in length and intakes were unrestricted. Two-bottle preference tests (30 min/day) were then conducted with half the rats being offered the CS1HF versus CS2 for 2 days, followed by the CS1HC versus CS2 for 2 days; the remaining rats were given the CS1HF and CS1HC in the reverse order. All rats were then given the choice between the CS1HF and the CS1HC for six 30-min/ day sessions. During the two-bottle tests, consumption of each CS remained paired with its respective i.g. infusion. To reduce side preferences, the left–right position of the CS solutions was alternated throughout training and testing. Food rations were given 3 h after the end of the sessions. Water was removed just prior to the sessions and returned along with the food rations. Data analysis. Intakes during training and preference sessions were averaged over 4-day (one-bottle) or 2-day (twobottle) blocks before being submitted to repeated-measures analyses of variance. Individual comparisons were evaluated by simple main effects or t-tests when appropriate. Intakes of the CS1s during two-bottle tests were also expressed as a percentage of total intake. Results and Discussion In the one-bottle training sessions, intakes of the CS1HF and CS1HC were very similar (6.7 and 6.9 mL/30 min), and
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES both were less (p , 0.05) than that of the CS2 (8.3 mL/30 min), F(2, 20) 5 4.7, p , 0.05. The rats were infused with similar amounts of the HF and HC diets (7.0 vs. 6.9 mL/session). During the two-bottle tests (Fig. 1), intake of both CS1s were greater than that of the CS2, F(1, 10) 5 11.8, p , 0.01. The difference was greater for the CS1HC versus CS2 than for the CS1HF versus CS2, but the CS1 type versus CS2 interaction failed to reach significance; the percent intake of the CS1HF was 72% and CS1HC was 60%, relative to the CS2. When given the choice between the two CS1s, the rats consumed more CS1HC than CS1HF over the six test sessions, F(1, 10) 5 6.6, p , 0.05. This preference decreased over repeated 2-day blocks but the CS by block interaction was not significant. The percent intake of CS1HC was 64% in the first two sessions and decreased to 51% by the last two sessions. This trend to decrease the CS1HC preference may have occurred because the rats were infused with both the HC and HF diets as they consumed the two CS1s in the choice tests. Thus, the CS1HC and CS1HF were no longer uniquely paired with their respective diet infusions. The pattern of results is close to what we found with isocaloric infusions of 7.1% corn oil and 16% maltodextrin using a similar paradigm (8). In that study, fat and carbohydrate were presented as pure sources, and the infusions were not as calorically dense as mixed diet infusions of the present experiment (0.64 kcal/mL vs. 2.1 kcal/mL). The carbohydrate and fat infusions conditioned CS1 preferences over the CS2 (73 and 68%, respectively), and the CS1C was preferred to the CS1F by 70%. (The CS1 preference test lasted two sessions, and thus, it is not known if the preference would have weakened with prolonged testing.) Taken together, the results suggest that i.g. carbohydrate is more effective than i.g. fat, whether presented in pure form or as part of a mixed diet, in conditioning flavor preferences when rats are food deprived and trained in short-term sessions.
391 EXPERIMENT 2
Whereas in Experiment 1 the food-deprived rats acquired a preference for the CS1HC over the CS1HF, just the opposite result was obtained in nondeprived rats trained with the same flavors paired with the same diet infusions during 22 h/ day sessions (4). These discrepant findings suggest that deprivation and/or session length significantly influences the relative reinforcing effects of high-fat and high-carbohydrate diets. This interpretation was verified in Experiment 2 by training the rats from the first experiment with the CS solutions during long-term sessions with food available ad lib. Methods Eight rats from Experiment 1 were used; the remaining rats developed problems with the i.g. cannula. After the last session of Experiment 1, the rats were given chow ad lib, and one-bottle training began 5 days later. The rats were given 12 one-bottle training sessions (22 h/day) with the CS1HC, CS1HF, and CS2 paired with their respective infusions; the CS flavors, i.g. infusions, and training sequence were the same as in Experiment 1. The rats were next given two-bottle preference tests (22 h/day) with the CS1HC versus CS2 (2 days) and CS1HF versus CS2 (2 days) with the order counterbalanced. This was followed by a CS1HC versus CS-HF preference test (6 days). Chow intake remained available ad lib, and intake was recorded to the nearest 0.1 g. CS drinking patterns (CS frequency and mean bout size) were analyzed using the licks/minute data and daily intake records. A bout was defined as a period of drinking containing a minimum of 10 licks and interlick intervals no larger than 10 min. Results and Discussion Table 2 summarizes the daily intakes and drinking patterns during the one-bottle training sessions. The rats differed in their total daily intakes of the CS solutions, F(2, 14) 5 15.7, p , 0.001; CS2 intake exceeded (p , 0.05) CS1HF intake which, in turn, was greater (p , 0.05) than CS1HC intake. These differences were due to variations in CS bout frequency, which mirrored those of total intakes, F(2, 14) 5 17.4, p , 0.001. CS bout sizes did not differ reliably. Because the rats drank more CS1HF than CS1HC, they also infused
TABLE 2 EXPERIMENT 2: MEANS 6 SEM OF CS AND ENERGY INTAKES AND CS BOUT PARAMETERS DURING 22-H ONE-BOTTLE TRAINING SESSIONS CS2
FIG. 1. Intakes (mean 1 SEM) of the CS solutions during the 30min/day two-bottle preference tests of Experiment 1. The CS1HC versus CS2 and CS1HF versus CS2 intakes represent the mean of the two sessions with each choice presented in a counterbalanced order. The CS1HC versus CS1HF intakes represent the means of test sessions 1 and 2 (left), 3 and 4 (middle), and 5 and 6 (right). CS1HC, CS1HF and CS2 were paired with intragastric infusions of HC diet, HF diet, and water, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
Total CS intake (ml/22 h) CS bout frequency (#/22 h) CS bout size (ml) IG diet (kcal/22 h) Chow intake (kcal/22 h) Total energy (kcal/22 h)
CS+HC
CS+HF
60.3 6 6.3‡
33.6 6 2.2*
44.7 6 3.1†
19.5 6 1.4‡ 3.1 6 0.2* —
11.4 6 0.7* 3.0 6 0.1* 73.9 6 5.2*
14.6 6 1.8† 3.3 6 0.2* 97.9 6 6.7†
55.9 6 1.6†
34.5 6 1.6*
35.6 6 1.0*
55.9 6 1.6‡
108.4 6 5.6*
133.5 6 6.2†
Data are 4-day means for each CS training condition. *†‡Within a row, means sharing a common superscript are not reliably different at the 0.05 level.
392 themselves with more HF diet than HC diet, t(7) 5 7.4, p , 0.001. Yet, the rats consumed similar amounts of chow on CS1HF days as on CS1HC days; chow intakes on CS1 days were reduced (p , 0.05), however, compared to CS2 days, F(2, 14) 5 76.3, p , 0.001. The rats consumed more (p , 0.05) total energy (chow 1 i.g. diet) on CS1HF days than CS1HC days, and more (p , 0.05) on both CS1 days than on CS2 days, F(2, 14) 5 133.7, p , 0.001. The results of the two-bottle choice tests are presented in Fig. 2. The rats consumed more CS1HF and CS1HC than the CS2, F(1, 7) 5 32.3, p , 0.001, but there was a reliable CS1 type by CS2 interaction, F(1, 7) 5 13.6, p , 0.01. Individual comparisons revealed a significant (p , 0.01) CS1HF versus CS2 difference but a marginal CS1HC versus CS2 difference (p 5 0.06); percent CS1HF and CS1HC intakes were 82 and 70%, respectively. Also, the intake of the CS1HF was greater (p , 0.01) than that of the CS1HC in these tests. When given the choice between the two CS1s, the rats consumed reliably more CS1HF than CS1HC, t(7) 5 2.6, p , 0.05; their percent CS1HF preference was 72%. The greater intake of the CS1HF than CS1HC resulted from an increased bout size [3.3 vs. 2.4 mL/bout, t(7) 5 4.6, p , 0.01] and a marginally increased bout number (8.6 vs. 4.1 bouts/day, p , 0.06). The two-bottle preference results are generally similar to those obtained in our prior study using the same HF and HC diets (4). The rats in the prior study showed slightly stronger preferences for the CS1HF (91%) and CS1HC (84%) relative to the CS2, and a somewhat weaker, but still reliable preference for the CS1HF over the CS1HC (64%). In both experiments, the rats took about twice as many CS1HF bouts as CS1HC bouts, but in contrast to the present results, bout sizes did not reliably differ in our earlier study (4). The one-bottle training results of the two studies also show a very similar pattern. In both cases, the rats consumed significantly more CS1HF than CS1HC by taking more CS1HF bouts/day, and they consumed more total energy on CS1HF
LUCAS AND SCLAFANI days than on CS1HC days: 23% more in this experiment, and 27% more in the prior experiment (4). These one-bottle training data would seem to offer an explanation for why the rats developed a CS1HF preference in this experiment, but not in Experiment 1. That is, in the first experiment training intakes of the two CS1s, and therefore, of the two diets were similar, but in the long-term sessions of the present experiment the rats consumed more CS1HF than CS1HC and thus were infused with more HF diet than HC diet. However, our earlier study (4) revealed that elevated intakes of the HF diet during training were not responsible for the CS1HF preference; rats developed a CS1HF preference even when intakes of the HF and HC diets were restricted to the same amount during training. The different preferences observed in Experiment 1, on the one hand, and the present experiment and prior study (4), on the other hand, would seem to be related to the different deprivation states (deprived versus nondeprived) or session durations (short-term versus long-term). EXPERIMENT 3
The third experiment examined whether the rats, after developing a CS1HF preference in Experiment 2, would continue to prefer this CS1 when retested during short-term sessions (30 min/day). The rats were tested under both fooddeprived and nondeprived conditions, the feeding states of Experiments 1 and 2, respectively. Methods Following the last choice test of Experiment 2, the rats (n 5 8) were placed on a food restriction schedule (40 kcal chow/day) and, 3 days later, were given a single 30-min session with 0.2% saccharin paired with i.g. water infusion; this was to prepare them for the short session duration. On the next 6 days they were given 30-min/day access to the CS1HC versus CS1HF. The rats were then returned to ad lib chow and, 4 days later, given another six sessions with CS1HF versus CS1HC available 30 min/day. Intakes of the two CS1s remained paired with their respective diet infusions. Chow was not available during the test sessions, and chow rations or ad lib chow were returned 3 h after the end of the sessions. Results and Discussion
FIG. 2. Intakes (mean 1 SEM) of the CS solutions during 22-h/day two-bottle preference tests of Experiment 2. The CS1HC versus CS2 and CS1HF versus CS2 intakes represent the mean of the two sessions with each choice presented in a counterbalanced order. The CS1HC versus CS1HF intakes represent the means of six test sessions. CS1HC, CS1HF, and CS2 were paired with intragastric infusions of HC diet, HF diet, and water, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
As illustrated in Fig. 3, the food-deprived rats consumed slightly, but not reliably more, CS1HC than CS1HF. The CS1HC preference was strongest during the first two session blocks (65%), and then declined to 56% by the last block, although the block by CS interaction failed to be significant (p 5 0.06; data not shown). In the nondeprived condition, intakes of the CS1HC and CS1HF were very similar and did not change over session blocks (Fig. 3). Despite showing a reliable CS1HF preference during the 22-h/day sessions of Experiment 3, the rats no longer preferred the CS1HF when food deprived and tested in 30-min sessions; in fact, they tended to prefer the CS1HC. They also showed no CS1 preference when tested nondeprived. Perhaps the rats had truly lost their preference for the CS1HF, or this preference may be expressed only during long-term choice sessions. The next experiment investigated these alternative explanations. EXPERIMENT 4
Beginning 3 days after the last test of Experiment 3, the rats were given a CS1HC versus CS1HF choice test, which
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES
393
FIG. 3. Intakes (mean 1 SEM) of the CS1 solutions during 30-min/ day two-bottle preference tests of Experiment 3. The intakes represent the means of six test sessions, each conducted under fooddeprived and food ad lib conditions. The C and CS1HF were paired with intragastric infusions of HC diet and HF diet, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
was both short term and long term. The CS1HC and CS1HF were presented for four 22-h access periods every other day. CS intakes and i.g. diet infusions during the first 0.5-h (shortterm test) and during the 0.5–22-h period (long-term test) were measured. CS1 solutions were presented every other day to make the short-term test phase comparable to the previous short-term tests, in which each daily 30-min session was preceded by 24 h without access to the CS1HC and CS1HF. Chow was not available during the short-term test (0.5 h), but was available during the long-term period (0.5–22 h). Results and Discussion In the short-term test (first 0.5 h) there was no difference in the intakes of the CS1HC and CS1HF (Fig. 4). However, in the long-term phase (0.5–22 h) the rats consumed reliably more CS1HF than CS1HC, t(7) 5 4.0, p , 0.01, which was due to both increased CS1HF bout frequency [6.2 vs. 3.4, t(7) 5 4.0, p , 0.01] and bout size [3.9 vs. 3.4 mL, t(7) 5 2.7, p , 0.05]. The percent CS1HF intakes in the short-term and long-term test periods were 44 and 67%, respectively. To further understand why the rats preferred the CS1HF in the long-term but not the short-term test period, the longterm period was divided into three intervals: 0.5–6 h, representing the period before lights out, and the first (7–12 h) and second half (13–18 h) of the dark period. Intakes during the early light period that followed the 12-h dark phase were very low (,1.0 mL), and were not included in the analysis. As indicated in Fig. 4, the difference between the intakes of the CS1HF and CS1HC increased over successive time periods, CS 3 period interaction, F(2, 14) 5 3.7, p , 0.05. During the light period (0.5–6 h) the rats did not consume reliably more CS1HF than CS1HC, but during both halves of the night CS1HF intakes exceeded (p , 0.05) CS1HC intakes, with the difference being slightly greater in the late night period. Given these results, it may be that the CS1HF preference varies as a function of the nycthemeral cycle, which could ex-
FIG. 4. Top: Intakes (mean 1 SEM) of the CS1 solutions during the first 30 min and remaining 21.5 h of the 22-h/day two-bottle preference tests of Experiment 4. Bottom: intakes (means 1 SEM) of the CS1 solutions during the 0.5–6 h (lights on), 6–12 h (lights off), and 12–8 h (lights off) periods of the 22-h/day two-bottle preference tests. CS1HC and CS1HF were paired with intragastric infusions of HC diet and HF diet, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
plain why the short-term choice tests (conducted during the light period) failed to reveal a CS1HF preference. To further evaluate nycthemeral variations in CS1 preference, we analyzed the CS1HF versus CS1HC data collected in our prior study (4) as a function of light/dark periods. The rats (n 5 11) in the second experiment of this prior study were trained and tested using the same CS solutions and diets as in the present study but only during 22 h/day sessions. The rats reliably preferred the CS1HF to the CS1HC over the 22 h/ day session, and their CS1HF preferences were very similar during the light period and during the first and second halves of the night (64–65%). These data suggest that the discrepancy in CS1HF preference between the short- and long-term sessions (present experiments) was not due to nycthemeral factors. In our prior study (4), the 22-h CS1HF versus CS1HC test sessions were conducted over successive days, whereas in the present experiment they were conducted on alternate days with the rats fed only low-fat chow on intervening days. It is possible that the rats in this experiment increased their CS1HF preference over the course of the 22-h sessions as
394
LUCAS AND SCLAFANI
they shifted from a low- to a high-fat intake. Prior work indicates that feeding rats a high-fat diet enhances fat appetite and fat-conditioned flavor preferences in some (7,11) but not all (8) cases. Further work is needed to resolve this issue. GENERAL DISCUSSION
This study sought to explain why opposite flavor preferences were obtained in prior work with rats trained with i.g. infusions of pure or mixed diets varying in fat/carbohydrate content (4,8,9). These studies differed in experimental conditions such as session duration and the animals’ deprivation state. The present data revealed that these parameters, rather than differences in the type of diet infused (pure versus mixed), account for the opposing results. When the rats were trained and tested in similar conditions (30-min/day sessions and food deprived), results obtained with mixed (Experiment 1) and pure (8,9) diets were in agreement: the rats preferred the high-carbohydrate (CS1HC or CS1C) over the high-fat (CS1HF or CS1F) paired flavor. In the present study, further experiments with the mixed diets evidenced the effect of training/testing conditions. The rats of Experiment 1 shifted from a high-carbohydrate to a high-fat preference when subsequently trained and tested in 22-h/day sessions, while nondeprived (Experiment 2). Experiments 3 and 4 further indicated that session length (30 min versus 22 h/day) influenced the expression of the fat- and carbohydrate-conditioned flavor preferences. Although the results of Experiment 1, obtained with mixed diets, are consistent with our previous results with pure macronutrients (8), the preference for the high-fat paired flavor in the present study was somewhat weaker (64 vs. 70%). One obvious reason for the smaller preference difference obtained with the mixed diets is that these diets were less distinct than the pure carbohydrate and fat sources; both the HF and HC diets contained fat (60 vs. 15% kcal), carbohydrate (34 vs. 79% kcal), and protein (6%). In addition, the studies differed in other respects, which may also contribute to the slightly differing results: the nutrient infusions were more calorically dense in the present study (2.1 vs. 0.64 kcal/mL), and the rats were infused with nutrient during preference testing in the present but not the prior study (8). In the present study, the preference for the high-fat versus high-carbohydrate paired flavors were tested in more sessions than in our prior work (six versus two sessions) (8). In two experiments (1 and 3) the preference tended to weaken with repeated testing; the reason for this is not clear. These details aside, the present findings along with those of two other recent studies (8,9) showed that in food-deprived rats, trained and tested in short-term sessions, the nutrient infusion with the higher carbohydrate content conditioned a stronger preference than the infusion with the higher fat content. This occurred even though the rats received as much [present experiment and (9)] or more energy (8) from the fat infusions as from the carbohydrate infusions. Perhaps the high-carbohydrate infusion was more reinforcing to fooddeprived rats than the high-fat infusion because the carbohydrate was more rapidly digested, absorbed, and utilized than the fat. This interpretation could be tested by using a more slowly processed carbohydrate (e.g., starch) and more rapidly processed fat (e.g., medium chain triglycerides) than used in the current research. The impact of deprivation state per se on the development and expression of fat- and carbohydrate-conditioned preferences remains an open question. In the short-term tests of Ex-
periment 3 the rats failed to show reliable flavor preferences under both deprived and nondeprived conditions, but these tests were conducted after the animals already had extensive experience with the flavors in short- and long-term sessions. To clarify this issue, separate groups of rats should be trained under deprived and nondeprived conditions using the shortterm conditioning procedure of Experiment 1. It is impractical to study the effects of deprivation state using the long-term conditioning procedure because the amount of nutrients infused during the sessions prevents energy deprivation. The long-term results obtained with the HF and HC diet infusions in Experiment 2 are very similar to those obtained in our prior study of these diets (4). In both studies, the rats selfinfused more HF than HC diet in the one-bottle training sessions, and preferred the CS1HF to the CS1HC in long-term choice tests. The new finding here is that after having acquired a CS1HF preference in the long-term sessions, the rats failed to prefer the CS1HF to the CS1HC during subsequent 30-min/day sessions (Experiment 3). They also consumed equivalent amounts of the two CS1s during the first 30 min of 22-h/day sessions, although they reliably preferred the CS1HF to the CS1HC during the remaining 21.5 h of the sessions (Experiment 4). It is possible that the rats’ initial 30-min/day training with the CS1s in the first experiment influenced their preference responses in the subsequent shortterm choice tests. Short-term preferences have not been measured in rats whose initial training was 22 h/day. As discussed in Experiment 4, it is also possible that the differences in the short-term and long-term CS1HF versus CS1HF results are related to nycthemeral variations in preferences or that cumulative effects of dietary fat intake affect preferences between low- and high-fat paired flavors. The present findings show that short-term preferences tests do not necessarily predict long-term food preferences and intake. As another example of this outcome, Lucas et al. (5) reported that rats preferred a flavor paired with 16% Polycose infusions over a flavor paired with 32% Polycose when food deprived and tested 30 min/day, but showed no preference when nondeprived and tested 22 h/day. The same study also showed that short- and long-term tests do not always disagree. Other rats trained with CS1 flavors paired with 8 and 16% Polycose infusions displayed strong preferences for the 16% paired flavor in short-term (deprived) and long-term (nondeprived) sessions. One common reason why short- and long-term preference tests may differ when animals are offered new food choices is that initial short-term preferences are influenced primarily by unlearned reactions to the new choices, while the long-term preferences reflect learned, as well as unlearned reactions to the now-familiar food choices [e.g., see (1)]. In Experiments 3 and 4 of the present study, however, the rats were quite familiar with the CS1HC and CS1HF flavors, and yet still showed different preference patterns in the short- and long-term tests. The relationship between short-term and long-term preference results is of importance to human feeding research where short-term food preference data are often collected to further our understanding of long-term preferences and intakes of low- and high-fat foods. ACKNOWLEDGEMENTS
This research was supported by a grant from the National Institute of Diabetes and Digestive and Kidney Diseases (DK-31135) and a National Institute of Mental Health Research Scientist Award (MH-00983) to Anthony Sclafani. The authors thank Dr. Karen Ackroff for her helpful comments on this article.
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES
395
REFERENCES 1. Ackroff, K.; Sclafani, A.: Sucrose to Polycose preference shifts in rats: The role of taste, osmolality and the fructose moiety. Physiol. Behav. 49:1047–1060; 1991. 2. Capaldi, E. D.: Conditioned food preferences. In: Capaldi, E. D., ed. Why we eat what we eat: The psychology of eating. Washington, DC: American Psychological Association; 1996:53–80. 3. Elizalde, G.; Sclafani, A.: Flavor preferences conditioned by intragastric Polycose: A detailed analysis using an electronic esophagus preparation. Physiol. Behav. 47:63–77; 1990. 4. Lucas, F.; Ackroff, K.; Sclafani, A.: High-fat diet preference and overeating mediated by postingestive factors in rats. Am. J. Physiol. 275:R1511–R1522; 1999. 5. Lucas, F.; Azzara, A. V.; Sclafani, A.: Flavor preferences conditioned by intragastric Polycose in rats: More concentrated Polycose is not always more reinforcing. Physiol. Behav. 63:7–14; 1997. 6. Lucas, F.; Sclafani, A.: Flavor preferences conditioned by intragastric fat infusions in rats. Physiol. Behav. 46:403–412; 1989.
7. Lucas, F.; Sclafani, A.: The composition of the maintenance diet alters flavor preferences conditioned by intragastric fat infusions in rats. Physiol. Behav. 60:1151–1157; 1996. 8. Lucas, F.; Sclafani, A.: Differential reinforcing and satiating effects of intragastric fat and carbohydrate infusions in rats. Physiol. Behav. 66:381–388; 1999. 9. Pérez, C.; Fanizza, L. J.; Sclafani, A.: Flavor preferences conditioned by intragastric nutrients in rats fed chow or a cafeteria diet. Appetite (in press). 10. Ramirez, I.: Stimulation of fluid intake by nutrients: Oil is less effective than carbohydrate. Am. J. Physiol. 272:R289–R293; 1997. 11. Reed, D. R.; Friedman, M. I.: Diet composition alters the acceptance of fat by rats. Appetite 14:219–230; 1990. 12. Sclafani, A.: How food preferences are learned—Laboratory animal models. Proc. Nutr. Soc. 54:419–427; 1995.
PII S0031-9384(98)00274-1
Flavor Preferences Conditioned by High-Fat versus High-Carbohydrate Diets Vary as a Function of Session Length FRANÇOIS LUCAS AND ANTHONY SCLAFANI1 Department of Psychology, Brooklyn College and the Graduate School, The City University of New York, 2900 Bedford Ave, Brooklyn, NY 11210-2889 Received 3 August 1998; Accepted 1 September 1998 LUCAS, F. AND A. SCLAFANI. Flavor preferences conditioned by high-fat versus high-carbohydrate diets vary as a function of session length. PHYSIOL BEHAV 66(3) 389–395, 1999.—Intragastric (i.g.) infusions of fat and carbohydrate condition flavor preferences in rats, but different results have been obtained in studies using pure and mixed nutrient infusions. This experiment compared the preference conditioning effects of mixed high-carbohydrate (HC) and high-fat (HF) diet infusions during short-term and long-term sessions. In Experiment 1 food-deprived rats were given one flavored saccharin solution (CS1HC) paired with i.g. infusions of an HC liquid diet, a second flavor (CS1HF) paired with HF diet infusions, and a third flavor (CS2) paired with i.g. water infusions during 30-min one-bottle training sessions. In subsequent two-bottle tests (30 min/day), the rats preferred both CS1s to the CS2 and preferred the CS1HC to the CS1HF. In Experiment 2, the same rats were trained and tested with the CSs and paired infusions during 22 h/day sessions with chow available ad lib. Both CS1s were again preferred to the CS2, but now the CS1HF was preferred to the CS1HC. When given additional 30-min choice sessions in Experiment 3 the rats showed no reliable preference for the CS1HC versus CS1HF under food-deprived or ad lib conditions. In Experiment 4, the rats were given 22-h CS1HC versus CS1HF choice sessions every other day. They showed no reliable CS preference during the first 30 min of each session, but reliably preferred the CS1HF during the remaining 21.5 h. These findings indicate that previously reported differences in preferences conditioned by pure versus mixed nutrient infusions are due to training procedures (session length, deprivation state) rather than to the type of nutrient infusions per se. The rats displayed different CS1HF versus CS1HC preferences as a function of test duration even after being given both shortand long-term training. Thus, short-term choice tests do not always predict the long-term intakes and preferences for high-fat and high-carbohydrate foods. © 1999 Elsevier Science Inc. Flavor preferences
Gastric infusions
Long-term test
Short-term test
In another recent study (4) rats were trained using complete liquid diets that differed in their fat and carbohydrate content. In this case, the rats acquired a preference for a flavor (CS1HF) paired with i.g. infusions of a high-fat (HF, 60% fat kcal) diet over another flavor (CS1HC) paired with infusions of a high-carbohydrate (HC, 79% CHO kcal) diet. It would appear from these data that fat is less reinforcing than carbohydrate when presented in a pure form, but more reinforcing when presented within a mixed diet. However, in addition to the type of nutrient infusions (pure versus mixed), the two studies differed in other respects, which may have contributed to the opposing results. In particular, the rats infused with pure nutrients were food deprived and trained in short (30-min) sessions, while the rats infused with mixed diets were nondeprived and trained in long (22-h) sessions. The
IT is now well established that food preferences in rats are conditioned by the postingestive reinforcing effects of nutrients (2,12). In recent studies (4,8,9) we have compared the reinforcing actions of fat (F) and carbohydrate (C) by pairing the intake of different flavored solutions (conditioned stimuli, CS) with intragastric (i.g.) infusions that differed in their fat/ carbohydrate content. As reported in the previous article (8), rats trained with one flavor (CS1F) paired with i.g. fat (7.1% corn oil) and another flavor (CS1C) paired with i.g. carbohydrate (16% maltodextrin) subsequently displayed a 70% preference for the CS1C over the CS1F [see also (9)]. Both flavors, however, were preferred to a third flavor (CS2), which was paired with i.g. water infusions. These findings confirm and extend earlier data indicating that fat infusions are less reinforcing than isocaloric carbohydrate infusions (3,6,7,10). 1To
Food deprivation
whom requests for reprints should be addressed. E-mail: [email protected]
389
390
LUCAS AND SCLAFANI
present experiment was conducted to determine whether deprivation state or session length influences the preference conditioning effects of high-fat and high-carbohydrate mixed diets.
EXPERIMENT 1
Methods Subjects. Eleven adult female Sprague–Dawley rats bred in our laboratory from Charles River Laboratories (Wilmington, MA) stock were used. The animals weighed 210–310 g at the beginning of the experiment. The animals were individually housed in wire-mesh cages in a room maintained at 218C under a 12:12-h light:dark cycle (lights on at 0800 h). Surgery. The rats were surgically implanted with stainless steel gastric cannulas as described by Elizalde and Sclafani (3). The animals were allowed to recover from surgery for at least 1 week prior to the start of training. When not in use, the cannulas were kept closed with stainless steel screws. Apparatus. Intragastric infusions were accomplished using an “electronic esophagus” apparatus described in detail elsewhere (3). In brief, the rats were housed in standard stainless steel cages modified so that powdered chow was available from a food cup accessible through a hole in the back wall of the cage. Drinking fluids were available through stainless steel sipper tubes available through two 19-mm holes in the front of the cage. Trays below the sipper tubes collected spillage. A slot in the cage floor permitted a stainless steel spring containing two catheters to connect the rat’s gastric cannula to outputs of a dual-channel infusion swivel located below the cage. Plastic tubing connected the swivel’s inputs to two peristaltic infusion pumps. The infusion pumps were operated automatically by drinkometer circuits and a microcomputer whenever the rat licked one or the other of two sipper tubes; the computer stored the number of licks/min for offline drinking pattern analysis. The rate of infusion was 1.5 mL/min, and about 1 mL of fluid was infused i.g. for each 1 mL of fluid orally consumed. Test solutions and diets. The CSs consisted of 0.2% sodium saccharin (Sigma, St. Louis, MO) solutions flavored with 0.05% cherry, grape or orange unsweetened Kool-Aid (General Foods, White Plains, NY). Unflavored 0.2% saccharin and 2% sucrose10.2% saccharin solutions were also used for training purposes. The i.g. infusions contained water, a high-fat (HF) diet, or a high-carbohydrate (HC) diet. The HF and HC diets were prepared by adding extra carbohydrate (maltodextrin ) and/or fat emulsion (corn oil) to evaporated milk, and they contained similar amounts of protein and energy (2.1 kcal/mL; see Table 1). The flavors paired with water, HC diet, and HF diet are referred to as CS2, CS1HC, and CS1HF, respectively. The specific flavor paired with each infusion was counterbalanced across subjects. Intakes of CS solutions and i.g. diets were recorded in g (60.1 g) and converted to milliliters. The rats were fed a maintenance diet of powdered Purina Chow (No. 5001). Procedure. Two weeks after surgery, the rats were familiarized with the saccharin solution by giving them two-bottle access to water and the saccharin1sucrose solution for 2 days followed by two-bottle access to water and saccharin for 2 days. They were then transferred to the test cages and attached to the infusion apparatus. The rats were maintained on a food restriction schedule that provided 40 kcal/day of chow. Total daily energy (chow 1 i.g. diet) intake averaged approximately 55 kcal/day throughout the experiment. This represented about 80% of their ad lib chow intake and was compa-
TABLE 1 COMPOSITION OF HIGH-FAT (HF) AND HIGH-CARBOHYDRATE (HC) DIETS
Evaporated milk,* g Maltodextrin,† g Corn oil,‡ g Sodium stearoyl lactylate,§ g Water, g Total, g Volume, ml kcal/ml % protein kcal % carbohydrate kcal % fat kcal
HF
HC
497.2 124.7 98.9 1.3 277.9 1000 855 2.1 6.4 34.0 59.6
458.2 320.1 — — 221.6 1000 934 2.1 6.4 79.0 14.6
*Pathmark Supermarket Brand. †Maltrin M580, Grain Processing Corp., Muscatine, IA. ‡Mazola, Best Foods, Englewood Cliffs, NJ. §Emplex, American Products, Kansas City, MO.
rable to the energy intakes allowed in our previous study of pure fat and carbohydrate conditioning (8). The rats were trained to drink the unflavored saccharin solution during seven daily 30-min sessions in the early light period (~2 h after light onset). During the last three of these pretraining sessions, intake of the saccharin solution was paired with i.g. water infusions. This was followed by one-bottle flavor conditioning sessions. Half of the rats were given the CS1HC paired with i.g. HC diet infusions on Day 1, the CS2 paired with i.g. water on Day 2, and the CS1HF paired with i.g. HF diet on Day 3. The remaining rats were given the CS1HF on Day 1 and the CS1HC on Day 3. This 3-day pattern was repeated three times for a total of 12 training sessions. The first three training sessions were 2 h in length to ensure that all rats would sample the CS solutions; oral and i.g. intakes were limited to a maximum of 10 mL each. The remaining training sessions were 30 min in length and intakes were unrestricted. Two-bottle preference tests (30 min/day) were then conducted with half the rats being offered the CS1HF versus CS2 for 2 days, followed by the CS1HC versus CS2 for 2 days; the remaining rats were given the CS1HF and CS1HC in the reverse order. All rats were then given the choice between the CS1HF and the CS1HC for six 30-min/ day sessions. During the two-bottle tests, consumption of each CS remained paired with its respective i.g. infusion. To reduce side preferences, the left–right position of the CS solutions was alternated throughout training and testing. Food rations were given 3 h after the end of the sessions. Water was removed just prior to the sessions and returned along with the food rations. Data analysis. Intakes during training and preference sessions were averaged over 4-day (one-bottle) or 2-day (twobottle) blocks before being submitted to repeated-measures analyses of variance. Individual comparisons were evaluated by simple main effects or t-tests when appropriate. Intakes of the CS1s during two-bottle tests were also expressed as a percentage of total intake. Results and Discussion In the one-bottle training sessions, intakes of the CS1HF and CS1HC were very similar (6.7 and 6.9 mL/30 min), and
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES both were less (p , 0.05) than that of the CS2 (8.3 mL/30 min), F(2, 20) 5 4.7, p , 0.05. The rats were infused with similar amounts of the HF and HC diets (7.0 vs. 6.9 mL/session). During the two-bottle tests (Fig. 1), intake of both CS1s were greater than that of the CS2, F(1, 10) 5 11.8, p , 0.01. The difference was greater for the CS1HC versus CS2 than for the CS1HF versus CS2, but the CS1 type versus CS2 interaction failed to reach significance; the percent intake of the CS1HF was 72% and CS1HC was 60%, relative to the CS2. When given the choice between the two CS1s, the rats consumed more CS1HC than CS1HF over the six test sessions, F(1, 10) 5 6.6, p , 0.05. This preference decreased over repeated 2-day blocks but the CS by block interaction was not significant. The percent intake of CS1HC was 64% in the first two sessions and decreased to 51% by the last two sessions. This trend to decrease the CS1HC preference may have occurred because the rats were infused with both the HC and HF diets as they consumed the two CS1s in the choice tests. Thus, the CS1HC and CS1HF were no longer uniquely paired with their respective diet infusions. The pattern of results is close to what we found with isocaloric infusions of 7.1% corn oil and 16% maltodextrin using a similar paradigm (8). In that study, fat and carbohydrate were presented as pure sources, and the infusions were not as calorically dense as mixed diet infusions of the present experiment (0.64 kcal/mL vs. 2.1 kcal/mL). The carbohydrate and fat infusions conditioned CS1 preferences over the CS2 (73 and 68%, respectively), and the CS1C was preferred to the CS1F by 70%. (The CS1 preference test lasted two sessions, and thus, it is not known if the preference would have weakened with prolonged testing.) Taken together, the results suggest that i.g. carbohydrate is more effective than i.g. fat, whether presented in pure form or as part of a mixed diet, in conditioning flavor preferences when rats are food deprived and trained in short-term sessions.
391 EXPERIMENT 2
Whereas in Experiment 1 the food-deprived rats acquired a preference for the CS1HC over the CS1HF, just the opposite result was obtained in nondeprived rats trained with the same flavors paired with the same diet infusions during 22 h/ day sessions (4). These discrepant findings suggest that deprivation and/or session length significantly influences the relative reinforcing effects of high-fat and high-carbohydrate diets. This interpretation was verified in Experiment 2 by training the rats from the first experiment with the CS solutions during long-term sessions with food available ad lib. Methods Eight rats from Experiment 1 were used; the remaining rats developed problems with the i.g. cannula. After the last session of Experiment 1, the rats were given chow ad lib, and one-bottle training began 5 days later. The rats were given 12 one-bottle training sessions (22 h/day) with the CS1HC, CS1HF, and CS2 paired with their respective infusions; the CS flavors, i.g. infusions, and training sequence were the same as in Experiment 1. The rats were next given two-bottle preference tests (22 h/day) with the CS1HC versus CS2 (2 days) and CS1HF versus CS2 (2 days) with the order counterbalanced. This was followed by a CS1HC versus CS-HF preference test (6 days). Chow intake remained available ad lib, and intake was recorded to the nearest 0.1 g. CS drinking patterns (CS frequency and mean bout size) were analyzed using the licks/minute data and daily intake records. A bout was defined as a period of drinking containing a minimum of 10 licks and interlick intervals no larger than 10 min. Results and Discussion Table 2 summarizes the daily intakes and drinking patterns during the one-bottle training sessions. The rats differed in their total daily intakes of the CS solutions, F(2, 14) 5 15.7, p , 0.001; CS2 intake exceeded (p , 0.05) CS1HF intake which, in turn, was greater (p , 0.05) than CS1HC intake. These differences were due to variations in CS bout frequency, which mirrored those of total intakes, F(2, 14) 5 17.4, p , 0.001. CS bout sizes did not differ reliably. Because the rats drank more CS1HF than CS1HC, they also infused
TABLE 2 EXPERIMENT 2: MEANS 6 SEM OF CS AND ENERGY INTAKES AND CS BOUT PARAMETERS DURING 22-H ONE-BOTTLE TRAINING SESSIONS CS2
FIG. 1. Intakes (mean 1 SEM) of the CS solutions during the 30min/day two-bottle preference tests of Experiment 1. The CS1HC versus CS2 and CS1HF versus CS2 intakes represent the mean of the two sessions with each choice presented in a counterbalanced order. The CS1HC versus CS1HF intakes represent the means of test sessions 1 and 2 (left), 3 and 4 (middle), and 5 and 6 (right). CS1HC, CS1HF and CS2 were paired with intragastric infusions of HC diet, HF diet, and water, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
Total CS intake (ml/22 h) CS bout frequency (#/22 h) CS bout size (ml) IG diet (kcal/22 h) Chow intake (kcal/22 h) Total energy (kcal/22 h)
CS+HC
CS+HF
60.3 6 6.3‡
33.6 6 2.2*
44.7 6 3.1†
19.5 6 1.4‡ 3.1 6 0.2* —
11.4 6 0.7* 3.0 6 0.1* 73.9 6 5.2*
14.6 6 1.8† 3.3 6 0.2* 97.9 6 6.7†
55.9 6 1.6†
34.5 6 1.6*
35.6 6 1.0*
55.9 6 1.6‡
108.4 6 5.6*
133.5 6 6.2†
Data are 4-day means for each CS training condition. *†‡Within a row, means sharing a common superscript are not reliably different at the 0.05 level.
392 themselves with more HF diet than HC diet, t(7) 5 7.4, p , 0.001. Yet, the rats consumed similar amounts of chow on CS1HF days as on CS1HC days; chow intakes on CS1 days were reduced (p , 0.05), however, compared to CS2 days, F(2, 14) 5 76.3, p , 0.001. The rats consumed more (p , 0.05) total energy (chow 1 i.g. diet) on CS1HF days than CS1HC days, and more (p , 0.05) on both CS1 days than on CS2 days, F(2, 14) 5 133.7, p , 0.001. The results of the two-bottle choice tests are presented in Fig. 2. The rats consumed more CS1HF and CS1HC than the CS2, F(1, 7) 5 32.3, p , 0.001, but there was a reliable CS1 type by CS2 interaction, F(1, 7) 5 13.6, p , 0.01. Individual comparisons revealed a significant (p , 0.01) CS1HF versus CS2 difference but a marginal CS1HC versus CS2 difference (p 5 0.06); percent CS1HF and CS1HC intakes were 82 and 70%, respectively. Also, the intake of the CS1HF was greater (p , 0.01) than that of the CS1HC in these tests. When given the choice between the two CS1s, the rats consumed reliably more CS1HF than CS1HC, t(7) 5 2.6, p , 0.05; their percent CS1HF preference was 72%. The greater intake of the CS1HF than CS1HC resulted from an increased bout size [3.3 vs. 2.4 mL/bout, t(7) 5 4.6, p , 0.01] and a marginally increased bout number (8.6 vs. 4.1 bouts/day, p , 0.06). The two-bottle preference results are generally similar to those obtained in our prior study using the same HF and HC diets (4). The rats in the prior study showed slightly stronger preferences for the CS1HF (91%) and CS1HC (84%) relative to the CS2, and a somewhat weaker, but still reliable preference for the CS1HF over the CS1HC (64%). In both experiments, the rats took about twice as many CS1HF bouts as CS1HC bouts, but in contrast to the present results, bout sizes did not reliably differ in our earlier study (4). The one-bottle training results of the two studies also show a very similar pattern. In both cases, the rats consumed significantly more CS1HF than CS1HC by taking more CS1HF bouts/day, and they consumed more total energy on CS1HF
LUCAS AND SCLAFANI days than on CS1HC days: 23% more in this experiment, and 27% more in the prior experiment (4). These one-bottle training data would seem to offer an explanation for why the rats developed a CS1HF preference in this experiment, but not in Experiment 1. That is, in the first experiment training intakes of the two CS1s, and therefore, of the two diets were similar, but in the long-term sessions of the present experiment the rats consumed more CS1HF than CS1HC and thus were infused with more HF diet than HC diet. However, our earlier study (4) revealed that elevated intakes of the HF diet during training were not responsible for the CS1HF preference; rats developed a CS1HF preference even when intakes of the HF and HC diets were restricted to the same amount during training. The different preferences observed in Experiment 1, on the one hand, and the present experiment and prior study (4), on the other hand, would seem to be related to the different deprivation states (deprived versus nondeprived) or session durations (short-term versus long-term). EXPERIMENT 3
The third experiment examined whether the rats, after developing a CS1HF preference in Experiment 2, would continue to prefer this CS1 when retested during short-term sessions (30 min/day). The rats were tested under both fooddeprived and nondeprived conditions, the feeding states of Experiments 1 and 2, respectively. Methods Following the last choice test of Experiment 2, the rats (n 5 8) were placed on a food restriction schedule (40 kcal chow/day) and, 3 days later, were given a single 30-min session with 0.2% saccharin paired with i.g. water infusion; this was to prepare them for the short session duration. On the next 6 days they were given 30-min/day access to the CS1HC versus CS1HF. The rats were then returned to ad lib chow and, 4 days later, given another six sessions with CS1HF versus CS1HC available 30 min/day. Intakes of the two CS1s remained paired with their respective diet infusions. Chow was not available during the test sessions, and chow rations or ad lib chow were returned 3 h after the end of the sessions. Results and Discussion
FIG. 2. Intakes (mean 1 SEM) of the CS solutions during 22-h/day two-bottle preference tests of Experiment 2. The CS1HC versus CS2 and CS1HF versus CS2 intakes represent the mean of the two sessions with each choice presented in a counterbalanced order. The CS1HC versus CS1HF intakes represent the means of six test sessions. CS1HC, CS1HF, and CS2 were paired with intragastric infusions of HC diet, HF diet, and water, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
As illustrated in Fig. 3, the food-deprived rats consumed slightly, but not reliably more, CS1HC than CS1HF. The CS1HC preference was strongest during the first two session blocks (65%), and then declined to 56% by the last block, although the block by CS interaction failed to be significant (p 5 0.06; data not shown). In the nondeprived condition, intakes of the CS1HC and CS1HF were very similar and did not change over session blocks (Fig. 3). Despite showing a reliable CS1HF preference during the 22-h/day sessions of Experiment 3, the rats no longer preferred the CS1HF when food deprived and tested in 30-min sessions; in fact, they tended to prefer the CS1HC. They also showed no CS1 preference when tested nondeprived. Perhaps the rats had truly lost their preference for the CS1HF, or this preference may be expressed only during long-term choice sessions. The next experiment investigated these alternative explanations. EXPERIMENT 4
Beginning 3 days after the last test of Experiment 3, the rats were given a CS1HC versus CS1HF choice test, which
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES
393
FIG. 3. Intakes (mean 1 SEM) of the CS1 solutions during 30-min/ day two-bottle preference tests of Experiment 3. The intakes represent the means of six test sessions, each conducted under fooddeprived and food ad lib conditions. The C and CS1HF were paired with intragastric infusions of HC diet and HF diet, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
was both short term and long term. The CS1HC and CS1HF were presented for four 22-h access periods every other day. CS intakes and i.g. diet infusions during the first 0.5-h (shortterm test) and during the 0.5–22-h period (long-term test) were measured. CS1 solutions were presented every other day to make the short-term test phase comparable to the previous short-term tests, in which each daily 30-min session was preceded by 24 h without access to the CS1HC and CS1HF. Chow was not available during the short-term test (0.5 h), but was available during the long-term period (0.5–22 h). Results and Discussion In the short-term test (first 0.5 h) there was no difference in the intakes of the CS1HC and CS1HF (Fig. 4). However, in the long-term phase (0.5–22 h) the rats consumed reliably more CS1HF than CS1HC, t(7) 5 4.0, p , 0.01, which was due to both increased CS1HF bout frequency [6.2 vs. 3.4, t(7) 5 4.0, p , 0.01] and bout size [3.9 vs. 3.4 mL, t(7) 5 2.7, p , 0.05]. The percent CS1HF intakes in the short-term and long-term test periods were 44 and 67%, respectively. To further understand why the rats preferred the CS1HF in the long-term but not the short-term test period, the longterm period was divided into three intervals: 0.5–6 h, representing the period before lights out, and the first (7–12 h) and second half (13–18 h) of the dark period. Intakes during the early light period that followed the 12-h dark phase were very low (,1.0 mL), and were not included in the analysis. As indicated in Fig. 4, the difference between the intakes of the CS1HF and CS1HC increased over successive time periods, CS 3 period interaction, F(2, 14) 5 3.7, p , 0.05. During the light period (0.5–6 h) the rats did not consume reliably more CS1HF than CS1HC, but during both halves of the night CS1HF intakes exceeded (p , 0.05) CS1HC intakes, with the difference being slightly greater in the late night period. Given these results, it may be that the CS1HF preference varies as a function of the nycthemeral cycle, which could ex-
FIG. 4. Top: Intakes (mean 1 SEM) of the CS1 solutions during the first 30 min and remaining 21.5 h of the 22-h/day two-bottle preference tests of Experiment 4. Bottom: intakes (means 1 SEM) of the CS1 solutions during the 0.5–6 h (lights on), 6–12 h (lights off), and 12–8 h (lights off) periods of the 22-h/day two-bottle preference tests. CS1HC and CS1HF were paired with intragastric infusions of HC diet and HF diet, respectively. Numbers atop bars represent the percentage of the corresponding CS solution.
plain why the short-term choice tests (conducted during the light period) failed to reveal a CS1HF preference. To further evaluate nycthemeral variations in CS1 preference, we analyzed the CS1HF versus CS1HC data collected in our prior study (4) as a function of light/dark periods. The rats (n 5 11) in the second experiment of this prior study were trained and tested using the same CS solutions and diets as in the present study but only during 22 h/day sessions. The rats reliably preferred the CS1HF to the CS1HC over the 22 h/ day session, and their CS1HF preferences were very similar during the light period and during the first and second halves of the night (64–65%). These data suggest that the discrepancy in CS1HF preference between the short- and long-term sessions (present experiments) was not due to nycthemeral factors. In our prior study (4), the 22-h CS1HF versus CS1HC test sessions were conducted over successive days, whereas in the present experiment they were conducted on alternate days with the rats fed only low-fat chow on intervening days. It is possible that the rats in this experiment increased their CS1HF preference over the course of the 22-h sessions as
394
LUCAS AND SCLAFANI
they shifted from a low- to a high-fat intake. Prior work indicates that feeding rats a high-fat diet enhances fat appetite and fat-conditioned flavor preferences in some (7,11) but not all (8) cases. Further work is needed to resolve this issue. GENERAL DISCUSSION
This study sought to explain why opposite flavor preferences were obtained in prior work with rats trained with i.g. infusions of pure or mixed diets varying in fat/carbohydrate content (4,8,9). These studies differed in experimental conditions such as session duration and the animals’ deprivation state. The present data revealed that these parameters, rather than differences in the type of diet infused (pure versus mixed), account for the opposing results. When the rats were trained and tested in similar conditions (30-min/day sessions and food deprived), results obtained with mixed (Experiment 1) and pure (8,9) diets were in agreement: the rats preferred the high-carbohydrate (CS1HC or CS1C) over the high-fat (CS1HF or CS1F) paired flavor. In the present study, further experiments with the mixed diets evidenced the effect of training/testing conditions. The rats of Experiment 1 shifted from a high-carbohydrate to a high-fat preference when subsequently trained and tested in 22-h/day sessions, while nondeprived (Experiment 2). Experiments 3 and 4 further indicated that session length (30 min versus 22 h/day) influenced the expression of the fat- and carbohydrate-conditioned flavor preferences. Although the results of Experiment 1, obtained with mixed diets, are consistent with our previous results with pure macronutrients (8), the preference for the high-fat paired flavor in the present study was somewhat weaker (64 vs. 70%). One obvious reason for the smaller preference difference obtained with the mixed diets is that these diets were less distinct than the pure carbohydrate and fat sources; both the HF and HC diets contained fat (60 vs. 15% kcal), carbohydrate (34 vs. 79% kcal), and protein (6%). In addition, the studies differed in other respects, which may also contribute to the slightly differing results: the nutrient infusions were more calorically dense in the present study (2.1 vs. 0.64 kcal/mL), and the rats were infused with nutrient during preference testing in the present but not the prior study (8). In the present study, the preference for the high-fat versus high-carbohydrate paired flavors were tested in more sessions than in our prior work (six versus two sessions) (8). In two experiments (1 and 3) the preference tended to weaken with repeated testing; the reason for this is not clear. These details aside, the present findings along with those of two other recent studies (8,9) showed that in food-deprived rats, trained and tested in short-term sessions, the nutrient infusion with the higher carbohydrate content conditioned a stronger preference than the infusion with the higher fat content. This occurred even though the rats received as much [present experiment and (9)] or more energy (8) from the fat infusions as from the carbohydrate infusions. Perhaps the high-carbohydrate infusion was more reinforcing to fooddeprived rats than the high-fat infusion because the carbohydrate was more rapidly digested, absorbed, and utilized than the fat. This interpretation could be tested by using a more slowly processed carbohydrate (e.g., starch) and more rapidly processed fat (e.g., medium chain triglycerides) than used in the current research. The impact of deprivation state per se on the development and expression of fat- and carbohydrate-conditioned preferences remains an open question. In the short-term tests of Ex-
periment 3 the rats failed to show reliable flavor preferences under both deprived and nondeprived conditions, but these tests were conducted after the animals already had extensive experience with the flavors in short- and long-term sessions. To clarify this issue, separate groups of rats should be trained under deprived and nondeprived conditions using the shortterm conditioning procedure of Experiment 1. It is impractical to study the effects of deprivation state using the long-term conditioning procedure because the amount of nutrients infused during the sessions prevents energy deprivation. The long-term results obtained with the HF and HC diet infusions in Experiment 2 are very similar to those obtained in our prior study of these diets (4). In both studies, the rats selfinfused more HF than HC diet in the one-bottle training sessions, and preferred the CS1HF to the CS1HC in long-term choice tests. The new finding here is that after having acquired a CS1HF preference in the long-term sessions, the rats failed to prefer the CS1HF to the CS1HC during subsequent 30-min/day sessions (Experiment 3). They also consumed equivalent amounts of the two CS1s during the first 30 min of 22-h/day sessions, although they reliably preferred the CS1HF to the CS1HC during the remaining 21.5 h of the sessions (Experiment 4). It is possible that the rats’ initial 30-min/day training with the CS1s in the first experiment influenced their preference responses in the subsequent shortterm choice tests. Short-term preferences have not been measured in rats whose initial training was 22 h/day. As discussed in Experiment 4, it is also possible that the differences in the short-term and long-term CS1HF versus CS1HF results are related to nycthemeral variations in preferences or that cumulative effects of dietary fat intake affect preferences between low- and high-fat paired flavors. The present findings show that short-term preferences tests do not necessarily predict long-term food preferences and intake. As another example of this outcome, Lucas et al. (5) reported that rats preferred a flavor paired with 16% Polycose infusions over a flavor paired with 32% Polycose when food deprived and tested 30 min/day, but showed no preference when nondeprived and tested 22 h/day. The same study also showed that short- and long-term tests do not always disagree. Other rats trained with CS1 flavors paired with 8 and 16% Polycose infusions displayed strong preferences for the 16% paired flavor in short-term (deprived) and long-term (nondeprived) sessions. One common reason why short- and long-term preference tests may differ when animals are offered new food choices is that initial short-term preferences are influenced primarily by unlearned reactions to the new choices, while the long-term preferences reflect learned, as well as unlearned reactions to the now-familiar food choices [e.g., see (1)]. In Experiments 3 and 4 of the present study, however, the rats were quite familiar with the CS1HC and CS1HF flavors, and yet still showed different preference patterns in the short- and long-term tests. The relationship between short-term and long-term preference results is of importance to human feeding research where short-term food preference data are often collected to further our understanding of long-term preferences and intakes of low- and high-fat foods. ACKNOWLEDGEMENTS
This research was supported by a grant from the National Institute of Diabetes and Digestive and Kidney Diseases (DK-31135) and a National Institute of Mental Health Research Scientist Award (MH-00983) to Anthony Sclafani. The authors thank Dr. Karen Ackroff for her helpful comments on this article.
HIGH-FAT AND HIGH-CARBOHYDRATE CONDITIONED PREFERENCES
395
REFERENCES 1. Ackroff, K.; Sclafani, A.: Sucrose to Polycose preference shifts in rats: The role of taste, osmolality and the fructose moiety. Physiol. Behav. 49:1047–1060; 1991. 2. Capaldi, E. D.: Conditioned food preferences. In: Capaldi, E. D., ed. Why we eat what we eat: The psychology of eating. Washington, DC: American Psychological Association; 1996:53–80. 3. Elizalde, G.; Sclafani, A.: Flavor preferences conditioned by intragastric Polycose: A detailed analysis using an electronic esophagus preparation. Physiol. Behav. 47:63–77; 1990. 4. Lucas, F.; Ackroff, K.; Sclafani, A.: High-fat diet preference and overeating mediated by postingestive factors in rats. Am. J. Physiol. 275:R1511–R1522; 1999. 5. Lucas, F.; Azzara, A. V.; Sclafani, A.: Flavor preferences conditioned by intragastric Polycose in rats: More concentrated Polycose is not always more reinforcing. Physiol. Behav. 63:7–14; 1997. 6. Lucas, F.; Sclafani, A.: Flavor preferences conditioned by intragastric fat infusions in rats. Physiol. Behav. 46:403–412; 1989.
7. Lucas, F.; Sclafani, A.: The composition of the maintenance diet alters flavor preferences conditioned by intragastric fat infusions in rats. Physiol. Behav. 60:1151–1157; 1996. 8. Lucas, F.; Sclafani, A.: Differential reinforcing and satiating effects of intragastric fat and carbohydrate infusions in rats. Physiol. Behav. 66:381–388; 1999. 9. Pérez, C.; Fanizza, L. J.; Sclafani, A.: Flavor preferences conditioned by intragastric nutrients in rats fed chow or a cafeteria diet. Appetite (in press). 10. Ramirez, I.: Stimulation of fluid intake by nutrients: Oil is less effective than carbohydrate. Am. J. Physiol. 272:R289–R293; 1997. 11. Reed, D. R.; Friedman, M. I.: Diet composition alters the acceptance of fat by rats. Appetite 14:219–230; 1990. 12. Sclafani, A.: How food preferences are learned—Laboratory animal models. Proc. Nutr. Soc. 54:419–427; 1995.