Substitutability of Nicotine and Sucrose in Rats: A Behavioral Economic Analysis

Abstract

Despite considerable evidence suggesting that sweet foods are a substitute for nicotine in humans, no formal behavioral economic analysis of this interrelationship has been conducted in nonhumans. The purpose of the present study was to examine this phenomenon in rats using concurrent schedules of sucrose pellet, chow pellet, and nicotine reinforcer delivery. Rats responded on separate levers that delivered sucrose pellets, chow pellets, or nicotine infusions under concurrent fixed-ratio (FR) 1 schedules for each commodity within a closed economy. Following stable food and nicotine intake, the unit price of either sucrose or nicotine (the primary commodity) was increased while the two alternative commodities remained unchanged. Substitution was quantified using a behavioral economic cross-price model, as well as a novel commodity relation index that normalizes consumption of dissimilar commodities. Asymmetrical partial substitution was observed, wherein sucrose served as a partial substitute for nicotine, but nicotine failed to substitute for sucrose. Moreover, sucrose was a stronger partial substitute for nicotine than chow in most rats. These findings indicate that substitution of food for nicotine depends on the type of food. These findings mirror the selective increase in carbohydrate intake that can occur during smoking cessation and demonstrate a behavioral economic mechanism that may mediate it.

1. Introduction

The impact of cigarette use on body weight is well established. Cigarette smoking reduces weight gain, while quitting smoking causes an individual to gain weight back to or above nonsmoker control levels (Albanes et al., 1987; Klesges et al., 1989). On average, smoking cessation leads to an increase in body weight of 4–5 kg within the first year, although up to 30% of ex-smokers gain more than 15 kg (Aubin et al., 2012; Flegal et al., 1995; Swan & Carmelli, 1995; Williamson et al., 1991). Human and animal research on how smoking and smoking cessation produces these effects shows that decreased body weight during smoking/nicotine delivery is due to reduced food intake (Bellinger et al., 2010; Grunberg et al., 1986; Miyata et al., 2001) and an increase in metabolic rate and physical activity (Hellerstein et al., 1994; Hofstetter et al., 1986; Hur et al., 2010; Jensen et al., 1995; Moffatt & Owens, 1991; Perkins et al., 1989). In contrast, increased body weight upon smoking cessation or nicotine withdrawal seems to be due primarily to increases in food intake (Grunberg et al., 1986; Miyata et al., 2001; Rodin, 1987; Stamford et al., 1986). Because weight gain is often a reason why people avoid quitting smoking or for relapse to smoking during a quit attempt (Camp et al., 1993; Fulkerson & French, 2003; Wiseman et al., 1998), understanding the mechanisms mediating the interactions between smoking and food intake may be helpful for developing more effective smoking cessation strategies.

The effects of smoking cessation on food intake appears to be specific to certain types of food, particularly those high in carbohydrates. Numerous studies have shown that calories from foods high in sugar and fat are increased during smoking cessation (Hall et al., 1989; Rodin, 1987; Spring et al., 1991; Wurtman & Wurtman, 1995), with post-cessation weight gain correlating highly with total carbohydrate intake (Rodin, 1987). Studies have also shown that motivation for carbohydrates increases during smoking cessation. For example, women who were abstinent from smoking for 24 hours performed computer tasks for carbohydrate snacks at lower probabilities of reinforcement compared to when they were not abstinent (Spring et al., 2003; see Lerman et al., 2004). Some studies in humans have shown that quitting smoking can also enhance the sensory effects (i.e., “pleasantness”) of sucrose solutions (Rodin, 1987). However, an increase in preference for higher sucrose solutions is not consistently observed, with some studies showing increased consumption of sweet solutions (Mullings et al., 2010; Pomerleau et al., 1991) and others showing a reduction (Redington, 1984).

The general appetite altering effects of nicotine and nicotine withdrawal are well documented in animal studies (e.g., Li et al., 2000; Grebenstein et al., 2013; Grunberg et al. 1986). In addition, we have previously shown that rats with concurrent access to sucrose, chow, and nicotine self-administration (NSA) exhibit an increase in sucrose intake during nicotine withdrawal (i.e., extinction of nicotine self-administration, Bunney et al., 2016). Specifically, although both chow and sucrose intake were increased during withdrawal, a larger percent increase in sucrose intake occurred relative to chow. Conversely, studies have shown that providing concurrent access to sucrose or other sweet alternative reinforcers (e.g., saccharin) reduces self-administration of nicotine and other drugs in rats (LeSage, 2009; Panlilio et al., 2015; Huynh et al., 2008).

These findings suggest that foods, especially those containing sugars, may be a behavioral economic substitute for nicotine/tobacco. According to operant behavioral economic theory, a reinforcer is defined as a substitute if its consumption concomitantly increases as consumption of an alternative reinforcer decreases with increasing price (Hursh, 1991). Specifically, this type of reinforcer interaction is typically demonstrated by constraining the consumption of one reinforcer via increasing its response cost (i.e., unit price), while the response cost of another reinforcer is held constant. Two human studies have examined the behavioral economic interaction between smoking and food intake (Epstein et al., 1991; Murphy et al., 2016). In Epstein et al., (1991), smokers were given a choice between puffs on a cigarette and a variety of food reinforcers (e.g., cheese, fruit, cookies, juice, etc.) under concurrent progressive variable-ratio (VR) schedules that increased concurrently for both alternatives with each reinforcer delivery (Epstein et al., 1991). When participants were nicotine deprived, they exclusively chose to smoke at all VR values. Interestingly, when food deprived, they preferred smoking at higher VR values, suggesting that smoking was an economic substitute for food. These findings suggest an asymmetrical economic relationship between smoking and food, wherein smoking serves as a substitute for food, but not vice versa. When overall changes in the consumption of specific food types were examined, both food and smoking deprivation increased fat intake, but not carbohydrate or protein intake. However, changes in the consumption of each type of food were not examined as a function of the price of smoking and food, making it unclear whether there was differential substitutability between food types. A study by Murphy et al., (2016) had participants make hypothetical choices between cigarettes and food (breakfast, lunch, dinner, and snack) options under a set budget. They found food and cigarettes functioned as independents rather than substitutes. However, a potential limitation of this study is that hypothetical choices may not accurately capture the impact of actual nicotine deprivation on real food consumption, as was the case with the Epstein et al., (1991) study. In addition, interactions between smoking and different types of food (e.g., fats versus carbohydrates) were not examined.

To our knowledge, there are no nonhuman studies that have examined the substitutability of sugar and nicotine using a formal behavioral economic framework. The purpose of the present study was to examine this issue by measuring the intake of each reinforcer at a fixed unit price as a function of increasing the unit price of the other concurrently available reinforcer. Because smoking and consumption of sweet food in humans occurs in the context of consuming other less-sweet types of food, we examined the substitutability of nicotine and sucrose while relatively bland lab chow was concurrently available at a fixed price.

2. Methods

2.1. Subjects

Sixteen naïve male Holtzman rats (Harlan, Indianapolis, IN) weighing 250–325 g at the start of the experiment were used in this study. All rats were individually housed in operant conditioning chambers in a temperature- and humidity-controlled room (20° C and 50%, respectively) with unlimited access to nicotine, 45 mg chow pellets (formula PJAI, TestDiet, Richmond, IN, USA; total kcal/g: 3.303; breakdown: 0.796 protein, 0.345 fat, and 2.162 carbohydrate [13% sucrose and 87% starch]), 45 mg sucrose pellets (formula PJFSC, TestDiet, Richmond, IN, USA; total kcal/g: 3.404; breakdown: 3.404 carbohydrate [99.998% sucrose and 0.002% starch]), and tap water under a reversed 12-h light/dark cycle (lights off at 11:00 h). Protocols were approved by the Institutional Animal Care and Use Committee of the Hennepin Healthcare Research Institute (formerly the Minneapolis Medical Research Foundation) in accordance with the 2013 NIH guide for the Care and Use of Mammals in Neuroscience and Behavioral Research.

2.2. Apparatus

Each operant conditioning chamber (29 cm × 26 cm × 33 cm; Coulbourn Instruments, Allentown, PA) was made of aluminum and Plexiglas walls, an aluminum ceiling, and a stainless-steel grid floor. Two standard response levers (ENV-110RM, Med Associates) were located on the front wall 7 cm above the chamber floor for chow and sucrose delivery and a third response lever was located on the back wall 7 cm above the chamber floor, for the delivery of nicotine. Standard grain pellets (45 mg chow pellets) were dispensed via a feeder (ENV-203M-45, Med Associates) into a food receptacle on the front wall located between two levers. Sucrose pellets (also 45 mg) were dispensed via a feeder into a food receptacle on the back wall between a water bottle and the third lever. Stimulus lights were located 2 cm above all three levers. A triple LED stimulus light was located above the nicotine lever, while a white stimulus light (one inch diameter) was located above the sucrose and chow levers. Water was continuously available via a spout mounted on the back wall of the chamber, to the left of the food receptacle. Each chamber was placed inside a sound-attenuating cubicle equipped with an exhaust fan that provided masking noise. Infusion pumps (Model RHSY, Fluid Metering, Syosset, NY) placed outside each cubicle delivered infusions through tygon tubing connected to a fluid swivel mounted above the chamber, and from the swivel through a spring leash connected to a guide cannula mounted in a harness assembly on the back of the rat. Med-PC IV (Med Associated, St Albans, VT) software was used for operating the apparatus and recording data.

2.3. Drugs

Nicotine bitartrate (Sigma Chemical Co., St. Louis, MO) was dissolved into sterile saline. The pH of the solution was adjusted to 7.4 with dilute NaOH, and heparin (30 units/ml) was added to help maintain catheter patency. Nicotine doses are expressed as the base. Methohexital (Sigma Chemical Co., St. Louis, MO) was dissolved in sterile saline (0.1ml, 10 mg/ml, i.v.) and used to assess catheter patency.

2.4. Surgery

Each rat was implanted with a chronic indwelling catheter into the right jugular or femoral vein, if the jugular catheter had failed, under ketamine (100.00 mg/kg, i.m.) / dexmedetomidine (0.25 mg/kg, i.m.) anesthesia, with atipamezole (1.00 mg/kg, s.c.) for reversal following surgery. The catheter was externalized between the scapulae and attached to a vascular-access harness (VAH95AB, Instech Laboratories, Plymouth Meeting, PA) that allowed connection to a fluid swivel via a tether for nicotine administration. Periodically, catheter patency was verified using methohexital (1.50 mg, i.v.). Catheter occlusion, as seen by failure to exhibit signs of anesthesia within 3–5 sec, resulted in the implantation of a new catheter in the left femoral vein in two rats.

2.5. Sucrose and Nicotine Demand Procedure

Rats were trained to lever press for nicotine, chow, and sucrose under concurrent FR1 schedules. At the beginning of each session, the stimulus light (green LED) above the nicotine lever was illuminated and each press on that lever resulted in offset of the stimulus light above the lever and i.v. delivery of nicotine (0.06 mg/kg/inf) in a volume of 100 μl/kg at a rate of 50 μl/sec, followed by a 7-sec timeout during which the stimulus light remained off and lever presses on the nicotine lever had no programmed consequence. Following the timeout, the stimulus light was illuminated, indicating availability of another nicotine infusion. For chow and sucrose delivery, each lever press on the respective lever resulted in immediate delivery of the relevant pellet and onset of the stimulus light above that lever for 1 sec. NSA was considered stable when a minimum of 10 infusions were earned per day with a coefficient of variance less than 15% and no obvious trend over five consecutive sessions. Chow and sucrose intake was considered stable when the number of each pellet earned exhibited no significant trend and <20% coefficient of variance over five consecutive sessions. An average of 41.3 (± 5.4 SEM) sessions was required for rats to reach stability. Six rats failed to acquire NSA and four rats that did acquire NSA became ill and could not complete the protocol, leaving a final sample size of seven rats.

After stable intake of all three reinforcers was achieved, rats were placed on an increasing FR schedule for either nicotine or sucrose, such that the FR was increased each session on the following schedule: 1, 2, 3, 6, 9, 15, 30, 60, 120, and doubling thereafter until the animal obtained 0 reinforcers on a particular FR requirement. This FR progression was used to better distinguish between relatively elastic and inelastic phases of consumption and improve accuracy of demand curve parameter estimates (see below). Then the FR was reduced back to FR 1. Once nicotine and sucrose intake were stable again, rats were placed on the same escalating FR schedule for the other reinforcer. Four out of seven animals received the FR escalation for nicotine first and the other three received the FR escalation for sucrose first. For both sucrose and nicotine, consumption reached zero at FR 30 for three rats and at FR 60 for the other four rats.

2.6. Data Analysis

2.6.1. Primary measures of nicotine, chow, and sucrose intake under FR escalation.

The mean number of nicotine infusions, chow pellets, and sucrose pellets per session over the last three sessions prior to FR escalation served as baseline. These baseline values and the number of infusions and pellets during each FR escalation session were used for demand curve analysis. Repeated measures ANOVA with Dunnett’s post tests was conducted to determine whether the number of pellets or infusions of each alternative reinforcer changed significantly from baseline. Data at each FR were also converted into a percentage of baseline, to examine the relative change in consumption of each alternative reinforcer. A repeated-measures mixed-effects model (reinforcer type x FR) with Sidak post tests was then performed (GraphPad Prism v.8.01) to compare the magnitude of percent change from baseline in the alternative reinforcers during nicotine or sucrose FR escalation (e.g., sucrose vs. chow during nicotine FR escalation, chow vs. nicotine during sucrose FR escalation).

2.6.2. Own-price exponential demand measurement.

The elasticity of own-price demand functions for nicotine and sucrose pellets represents the elasticity of demand of these reinforcers as the unit price (FR) was increased under conditions when fixed-price alternatives were available (e.g., sucrose own-price demand had nicotine and chow concurrently available on a FR 1). This was quantified by fitting the total reinforcer deliveries across FR values for each rat with the (Hursh & Silberberg, 2008) exponential demand equation:

$$\log Q=\text{log}{Q}_0+k\left(e^{-\alpha Q_{0}C}-1\right)$$ (1)

where Q is the quantity of a primary commodity consumed (reinforcers earned), C is unit-price cost of the primary commodity (FR/reinforcer), and Q₀ and α are best-fit free parameters that refer to maximal consumption at zero price and rate of change in consumption across price (i.e., demand elasticity), respectively. The scaling parameter, k, is a constant that is fit globally across individuals/conditions to normalize the range of consumption across commodities. Such normalization allows for comparisons of free parameter estimates (i.e., α and Q₀) of individual subjects between the different demand functions using a dependent-samples t-test (note: α values were log transformed prior to conducting statistical analysis due to non-normality of the distribution). To provide a complete demand function, unit prices where consumption was 0 were replaced with 0.1 since 0 is undefined on a log scale and the log of 0.1 (i.e., log 0.1 = −1) is the next lowest log-unit value below the log of 1 infusion (i.e., log 1 = 0). Additionally, to include all rats at all unit prices in the mean group fits of demand functions, 0 infusions (i.e., 0.1) were interpolated for each subject from the point where 0 infusions were earned to the highest unit price achieved by any individual rat. These interpolated data were not used to determine the individual-subject fits of demand curves and were only used to portray group demand functions. Values of P_max (the point where consumption changes from inelastic to elastic) and O_max (the maximum response output) were determined from double-normalized demand curves to allow comparison across the different commodity types. To accomplish this, consumption was expressed as the percentage of consumption at Q₀ and price was expressed as the number of responses to obtain 1% of maximal consumption.

2.6.3. Cross-price demand measurement.

The cross-price elasticity of nicotine, sucrose, and chow represents the change in intake when each was available as a fixed-price alternative. This was quantified by fitting a curve to the mean number of fixed-price reinforcer deliveries across the unit price of the own-price alternative (e.g., the number of nicotine or chow reinforcers delivered as the unit price [FR] of sucrose was increased) using the (Hursh & Roma, 2013) cross-price demand equation:

$$Q=\text{log}{Q}_{\mathit{\text{alone}}}+I{e}^{-\beta \text{C}}$$ (2)

where Q is consumption of the fixed-price commodity, Q_alone is consumption of that commodity at the highest unit-price, I is the interaction constant (i.e., the log-unit change in consumption), β is the price sensitivity (i.e., rate of change) in consumption of the fixed-price commodity to changes in the primary commodity price, and C is the unit-price of the primary commodity. For three fits, the best-fit Q_alone values were inaccurate (~3 fold greater than observed intake at the highest price of the primary commodity) due to the near linear fits of the cross-price functions. To correct for this, Q _alone values were fixed at the average consumption from the two highest prices of the primary commodity, which resulted in better r² values than just using consumption at the highest price. Significance of substitution parameters β and I was judged via one-sample t-tests with a hypothetical value of zero and p < 0.05. To compare individually fit βand I values between conditions, dependent-samples t-tests were conducted (p < 0.05) following log transformation to normalize the distribution of these measures.

2.6.4. Normalized Commodity Relation Index.

The utility of I to quantify the degree of substitutability or complementarity of an alternative commodity is limited since it expresses the log-unit change in consumption, not the absolute change in consumption (i.e., consumption increasing from 1 to 10 or from 100 to 1000 would both be expressed as an I = −1.0). As such, I obscures the absolute magnitude of difference in the amount of substitution observed. Previously, we proposed a Commodity Relation Index (CRI; see (Smethells et al., 2018)) that provides a metric to quantify both the nature (e.g., substitute, complement) and degree (e.g., partial, full) of the relationship between two commodities as the proportional change in consumption of an alternative commodity over that of the primary commodity (i.e., Δ alternative/Δ primary). The CRI, however, has one important limitation: reinforcers without a common unit of measurement (e.g., sucrose pellets vs nicotine infusions) cannot be reciprocally compared between cross-price demand conditions to determine if, for example, the degree of substitution is equivalent between them (e.g., symmetrical vs asymmetrical substitution), since reinforcer consumption is not normalized (e.g., mg/kg of nicotine). As a solution, we propose an alternative version of this CRI equation termed the Normalized Commodity Relation Index (CRI_n) that quantifies the proportional change in consumption of the alternative commodity relative to its own maximal consumption. Such normalization allows for a comparison of the relative change in consumption between reinforcers that have a different unit of measurement. To calculate the CRI_n, the demand intensity (y-intercept), termed Q_{0 cross}, of the fixed-price alternative needs to be calculated using Q_alone and I:

$$Q_{0\mathit{\text{cross}}}={10}^{\mathit{\text{log}}\left(Q_{\mathit{\text{alone}}}\right)+I}$$ (3)

and once calculated, it is incorporated with Q_alone to determine the CRI_n:

$$CR{I}_n=\frac{\left(Q_{\mathit{\text{alone}}}-Q_{0\mathit{\text{cross}}}\right)}{\frac{1}{2}\left(Q_{0\mathit{\text{cross}}}+Q_{\mathit{\text{alone}}}+\left|Q_{0\mathit{\text{cross}}}-Q_{\mathit{\text{alone}}}\right|\right)}$$ (4)

where positive CRI_n values indicate the alternative functions as a complement, negative CRI_n values indicate it functions as a substitute, and values close to 0 indicate it functions as an independent. The denominator of the CRI_n is a mathematical formula to determine the maximum of Q_alone and Q_{0 cross} (note the absolute value brackets in equation 4), which normalizes the change in consumption relative to the maximum consumption observed to produce comparable positive and negative CRI_n values between −1 and 1 when consumption increases and decreases, respectively (e.g., (8–2)/8 = 0.75 vs. (2−8)/8 = −0.75). The constrained range of this index also allows it to provide a broader description of the strength of the relationship from weak (0.1 – 0.3) to moderate (0.3 – 0.6) to strong (0.6 – 1.0). Although a visual/graphic inspection of the data can easily convey the nature and degree of interaction between reinforcers, it does not provide a quantitative index of the interaction. The CRI_n is a quantitative measure that allows scaling interactions between different pairs of reinforcers, or the same pair of reinforcers between subjects or experimental conditions. Significance of substitution (CRI_n) was judged via one-sample t-tests with a hypothetical value of zero and p < 0.05. Comparison of the CRI_n between conditions was made via paired t-test, with p < 0.05.

3. Results

3.1. Own-price demand analysis.

To quantify the economic relationship between nicotine and sucrose across subjects, an analysis was conducted on reinforcer consumption across unit price using the best-fit own-price demand parameters (see Table 1). Figures 1 and 2 represent the own-price exponential demand curve fits (left panels) for sucrose and nicotine respectively. Own-price demand elasticity was more inelastic for sucrose (Figure 1 – open circles) than it was for nicotine (Figure 2 – closed triangles), which was indicated by significantly smaller alpha values (t₆ = 2.858, p < 0.05) and significantly greater P_max (t₆ = 2.853, p < 0.05) and O_max (t₆ = 2.282, p < 0.05) values for sucrose compared to nicotine (see Table 1 for parameter values).

Table 1.

Exponential demand curve parameters for individual rats (Note: P_max and O_max values are derived from double-normalized demand functions to allow comparisons amongst parameters - see Own-Price Exponential Demand Measurement in the Methods section)

Sucrose Own-price Demand						Nicotine Own-price Demand
ID #	Q0	α	Pmax	Omax	R2	ID #	Q0	α	Pmax	Omax	R2
2424	181.8	0.000096	12.48	310.34	0.97	2424	53.8	0.000199	5.14	176.56	0.99
2514	159.0	0.000431	2.24	78.40	0.97	2514	49.7	0.000703	1.02	80.76	0.96
2443	337.9	0.000101	13.23	290.00	0.99	2443	14.6	0.00186	0.14	42.88	0.99
2502	42.52	0.000522	2.19	86.92	0.90	2502	11.9	0.002098	0.09	39.41	0.99
2445	88.62	0.000478	1.86	98.65	0.99	2445	18.8	0.000881	0.41	60.34	0.89
2512	121.7	0.000414	2.41	99.27	0.99	2512	18.0	0.000834	0.63	63.06	0.98
2448	239.9	0.000105	12.81	326.47	0.95	2448	11.6	0.000588	1.65	64.91	0.91
Mean	167.4	0.000307 *	6.75 *	184.30 *	0.97	Mean	25.5	0.001023	1.30	75.42	0.96
SEM	37.3	0.000074	2.16	44.33	0.01	SEM	6.9	0.000262	0.67	17.66	0.02

^*Different from nicotine, p < 0.05

Figure 1.

Individual and group mean (± S.E.M.) reinforcer deliveries across the unit price of sucrose during the sucrose own-price demand and the nicotine and chow cross-price demand assessments (see Table 1 – left column – for equation parameter values).

Figure 2.

Individual and group mean (± S.E.M.) reinforcer deliveries across the unit price of nicotine during the nicotine own-price demand and the sucrose and chow cross-price demand assessments (see Table 1 – right column – for equation parameter values).

3.2. Cross-price demand analysis using the CRI_n.

Figure 3 shows the CRI_n values for the alternative commodities during Own-price demand for sucrose (left panel) and nicotine (right panel). Across these conditions, the alternative commodity that emerged as the dominate substitute (higher CRI_n) differed, indicating asymmetrical substitutability between nicotine and sucrose (see Table 2 for Cross-price Demand parameter values). Specifically, during the sucrose own-price demand assessment, the chow CRI_n was significantly greater than 0 (95% CI: 0.14 – 0.52) and significantly higher than the nicotine CRI_n (t₆ = 3.287, p < 0.05), which did not differ from 0 (95% CI: −0.29 – 0.38). Thus, chow, but not nicotine, served as a substitute for sucrose. During nicotine own-price demand assessment, the sucrose CRI_n was significantly greater than 0 (95% CI: 0.31 – 0.68) and was significantly higher than the chow CRI_n (t₆ = 3.032, p < 0.05), which did not differ from 0 (95% CI: −0.09 – 0.31). Thus, sucrose, but not chow, served as a substitute for nicotine. Because the sucrose CRI_n was less than 1, sucrose was only a partial substitute for nicotine. Furthermore, between the two own-price demand assessments, the CRI_n for sucrose was significantly greater than that for nicotine (t₆ = 2.691, p < 0.05), further indicating an asymmetrical substitution between these two commodities. The β parameter (i.e, the initial rate of change in consumption of the alternative reinforcer), however, was not significantly different between sucrose and nicotine, indicating that the initial slopes of substitution as unit price increased were similar.

Figure 3.

Normalized Commodity Relation Index (CRI_n) values for chow, sucrose and nicotine during the sucrose own-price demand (left panel) and nicotine own-price demand (right panel) conditions. * Significant difference (p < 0.05) between the two alterative reinforcer commodities.

Table 2.

Cross-price demand curve parameters for individual rats during sucrose (left columns) and nicotine (right columns) demand assessment.

Nicotine Cross-price Demand							Sucrose Cross-price Demand
ID #	Q0 cross	β	I	Qalone	CRI n	R2	ID #	Q0 cross	β	I	Qalone	CRI n	R2
2424	17.2	1.000	−0.33	36.6	0.53	0.3	2424	65.8	1.000	−0.57	246.0	0.73	0.61
2514	31.5	0.219	0.08	26.2	−0.17	0.28	2514	16.0	0.020	−0.46	46.0	0.65	0.20
2443	7.5	0.794	−0.25	13.1	0.43	0.22	2443	244.4	0.086	−0.09	298.1	0.18	0.60
2502	13.8	0.045	0.12	10.6	−0.23	0.43	2502	107.7	0.480	−0.32	227.5	0.53	0.88
2445	36.8	0.526	0.19	23.6	−0.36	0.25	2445	61.0	0.034	−0.22	101.9	0.40	0.59
2512	26.0	0.277	0.08	21.5	−0.17	0.34	2512	38.3	0.060	−0.44	105.3	0.64	0.74
2448	18.5	0.318	−0.15	26.3	0.30	0.12	2448	198.1	0.082	−0.18	302.8	0.35	0.83
Mean	21.6	0.454	−0.04	22.6	0.05 * ‡	0.28	Mean	104.5	0.252	−0.33++	189.7	0.5 ++, * ,#	0.64
SEM	3.89	0.128	0.08	3.3	0.14	0.04	SEM	32.3	0.139	0.07	39.2	0.07	0.08
Chow Cross-price Demand							Chow Cross-price Demand
ID #	Q0 cross	β	I	Qalone	CRI n	R2	ID #	Q0 cross	β	I	Qalone	CRI n	R2
2424	99.2	0.451	−0.51	322.4	0.69	0.91	2424	266.9	1.000	0.10	211.8	−0.21	0.11
2514	247.5	0.315	−0.23	424.7	0.42	0.75	2514	394.2	0.136	−0.23	669	0.41	0.95
2443	291.0	0.223	−0.24	503.2	0.42	0.86	2443	230.1	0.144	−0.06	265.5	0.13	0.49
2502	405.7	0.130	−0.09	500.8	0.19	0.70	2502	350.3	0.186	0.00	353.3	0.01	0.01
2445	422.3	0.486	−0.06	489.3	0.14	0.50	2445	320.8	0.515	−0.18	489.0	0.34	0.75
2512	395.0	1.000	−0.05	441.6	0.11	0.04	2512	518.1	1.000	0.03	481.6	−0.07	0.01
2448	325.1	0.173	−0.19	497.8	0.35	0.88	2448	330.8	0.012	−0.06	383.8	0.14	0.26
Mean	312.3	0.397 +	−0.2+	454.3	0.33 ++,‡	0.66	Mean	344.5	0.428	−0.06	407.7	0.11 #	0.37
SEM	43.06	0.113	0.06	24.9	0.08	0.12	SEM	35.4	0.159	0.04	58.2	0.08	0.14

Significantly different from zero, ⁺p < 0.05, ⁺⁺p < 0.01, ⁺⁺⁺p < 0.001.

Significant (p < 0.05) differences in best-fit parameters for the cross-price demand equation are denoted by different symbols for comparisons between Nicotine and Sucrose (*), Nicotine and Chow (ｷ) and Sucrose and Chow (#).

3.3. Group mean change in consumption of alternative reinforcers during FR escalation.

Figure 4 shows the group mean reinforcer consumption during FR escalation as absolute values (upper panels) and a percentage of baseline (lower panels). When the change in chow and nicotine intake during FR escalation for sucrose was examined (Figure 4, upper left panel), absolute chow consumption increased as sucrose consumption decreased (F_{1.1, 5.9} = 9.05; p < 0.05), with consumption significantly higher than baseline at sucrose FR 6 to FR30, and maintaining a trend upward at FR 60. In contrast, there was no significant change in absolute nicotine intake at any unit price for sucrose. When the percent change in chow and nicotine intake were compared (Figure 4, lower left panel), there were significant main effects of alternative reinforcer type (F_{1, 6} = 17.26; p < 0.01) and reinforcer type by FR interaction (F_{7, 36} = 7.17; p < 0.001), but not FR. The percent increase in chow consumption was significantly greater than that for nicotine at FR 9, 15, and 30, but not FR 60. Three rats were not exposed to sucrose FR 60 because they had reached zero sucrose consumption at FR 30. During FR escalation for nicotine (Figure 4, upper right panel), absolute sucrose consumption increased (F_{2.3, 12.8} = 6.98; p < 0.01), with intake higher than baseline at nicotine FR 30 and 60. In contrast, there was no significant change in chow intake. When percent change in chow and sucrose intake during FR escalation for nicotine was compared, there was no effect of alternative reinforcer type, but a significant main effect of FR (F_{1.0, 6.2} = 6.19; p < 0.05) and the reinforcer type by FR interaction approached significance (F_{1.7, 8.6} = 3.717; p = 0.07). Comparisons between chow and sucrose at each nicotine FR were not significant. Three rats were not exposed to nicotine FR 60 because they had reached zero nicotine intake at FR 30.

Figure 4.

Mean group consumption of chow, sucrose and nicotine expressed as absolute values (top panels) and a percentage of baseline consumption (FR 1, bottom panels) across the primary commodity unit price (left: sucrose; right: nicotine). Each point is the mean of seven rats, except at FR 60, which is the mean of four rats. Difference in absolute reinforcers from FR 1 for each alternative reinforcer denoted by: * = (p < 0.05); ** = (p > 0.01). Significant difference in percent baseline between the two alterative reinforcers denoted by: # = (p < 0.05).

4. Discussion

The purpose of the present study was to formally examine the substitutability between sucrose pellets (~100% sucrose) and nicotine under concurrent access conditions using a behavioral economic analysis. A relatively starchy food (chow – 87% starch and 13% sucrose content) was also concurrently available to determine whether the substitutability of a food reinforcer for nicotine is influenced by the type (i.e., macronutrient content) of food available. The primary finding was an asymmetric substitution between sucrose and nicotine, with sucrose serving as a partial substitute for nicotine, but nicotine failing to substitute for sucrose in most rats. Moreover, sucrose was a stronger substitute for nicotine than chow in most rats, indicating that the substitution of food for nicotine depends on the form/concentration of carbohydrate in the food reinforcer. To our knowledge, this is the first nonhuman study to formally examine substitution between nicotine and sucrose using a behavioral economic analysis, and the findings make important contributions to a) understanding the relationship between smoking and food motivation, b) the experimental analysis of relative reinforcing efficacy and interactions between dissimilar reinforcers, c) the quantitative approaches to measuring those interactions, and d) anticipating the potential impact of individual and population changes in tobacco use on food consumption and associated health concerns (e.g., obesity).

The lack of a significant increase in chow intake as nicotine intake decreased in the present study differs from the findings of Bunney et al., (2016), where significant increases in both chow and sucrose intake occurred during extinction of nicotine self-administration (Bunney et al., 2016). However, the greater relative increase in sucrose intake compared to chow intake in that study is consistent with the selective substitution of sucrose for nicotine in the present study. In addition, the present findings are consistent with those of Grunberg et al. (1985) showing a selective increase in simple sugar compared to complex carbohydrate consumption during withdrawal from continuous nicotine infusion. The lack of change in chow consumption in the current study may have been due to the relatively gradual decline in nicotine intake and brevity of nicotine demand assessment (6–8 days) compared to the immediate withdrawal of nicotine and longer duration of extinction in the Bunney et al. study (10–16 days to stability). Nonetheless, these studies are consistent in showing that decreased nicotine intake produces a greater increase in motivation for foods that are relatively high in simple sugar content compared to those mostly composed of complex carbohydrates. Results of the current study are also consistent with human studies showing that a) quitting smoking is associated with increased caloric intake in the form of high sugar and high fat foods (Hall et al., 1989; Rodin, 1987; Spring et al., 1991; Wurtman & Wurtman, 1995), b) carbohydrate intake is positively correlated with withdrawal symptoms during smoking abstinence (Ogden, 1994; Pepino & Mennella, 2007; Spring et al., 2003), and c) withdrawal from nicotine enhances motivation for snacks high in simple sugars (Spring et al., 2003). Together, these human studies suggest that sweet foods serve as an economic substitute for smoking (Bunney et al., 2016; Epstein et al., 1991). The present study formally demonstrates this behavioral mechanism and that sucrose is an imperfect substitute for nicotine.

However, the present findings contrast with those from Epstein et al., (1991), in which smoking deprivation produced an increase in consumption of fat, but not carbohydrates or protein. This discrepancy could be due to differences in how smoking/nicotine intake was constrained (i.e., by deprivation in Epstein et al. vs differential price manipulation in the present study). Also, several types of food were available from three different classes (high-simple sugar, high-fat, high-protein) in Epstein et al., whereas only sugar alone and a nutrient-balanced food were available in the present study. The substitutability of sucrose for nicotine in rats may depend upon concurrent availability of other high-macronutrient foods not used in the present study (e.g., high-fat food pellets). However, a limitation of the present study and Epstein et al. is that foods varied in terms of flavors and other sensory factors that may have influenced intake independent of the primary macronutrient present in each item. More research on the interaction between flavors, sensory factors, and macronutrient content between different food types is needed to address this issue.

The own- and cross-price behavioral economic models proved useful for quantifying the relative reinforcing efficacy of sucrose and nicotine and the nature of the economic relationship between these reinforcers within the context of concurrently available chow. The present results are consistent with previous work by Panlilio et al., (2015) which showed that sucrose pellets are preferred to concurrently available nicotine under similar training conditions, suggesting that the reinforcing efficacy of sucrose is greater than nicotine. The present findings are also reminiscent of another study showing that demand for chocolate-flavored food pellets was greater than that for nicotine when they were not concurrently available under open economies (i.e., each was examined in separate experiments; Chellian et al., 2019). The present study extends this work by showing that the greater reinforcing efficacy of sucrose compared to nicotine is evident in terms of elasticity of demand under concurrent access and closed-economy conditions, as well as preference. In addition, the present study demonstrates the robustness of this difference in reinforcing efficacy, in that, despite the availability of another fixed-price food source that would be expected to reduce motivation for sucrose to some degree, elasticity of demand for sucrose was still greater than nicotine.

The present study extends the broader literature on the economic relationship between concurrently available drug and nondrug reinforcers. The ability of a nondrug reinforcer to substitute for a drug reinforcer depends upon the type of reinforcer, drug reinforcer dose, level of food deprivation, and the nature of the economy in which each reinforcer is available (Bickel et al., 1995; Foltin, 1997, 1999). In the present study, sessions were long (23 hr) and a closed economy was used for both food reinforcers, in which no consumption of either reinforcer was allowed outside the experimental sessions. Studies using a similar approach have shown that ethanol and cocaine can serve as an economic substitute for food pellets in baboons (Foltin, 1998, 1999). Amphetamine can also substitute for food pellets in baboons, but only when access to food is restricted (Foltin, 1997). Studies using relatively short sessions and an open economy for food also show that ethanol substitutes for sucrose (Petry & Heyman, 1994; Samson et al., 1982). Findings on the symmetry of substitution between drug and food reinforcers have been mixed, with food substituting for amphetamine (Foltin, 1997), but sucrose failing to substitute for ethanol (Foltin, 1998). In addition, Dworkin et al., (1990) also showed that food failed to substitute for morphine, although symmetry per se was not examined. The present study extends this literature by showing that nicotine and sucrose also exhibit an asymmetrical substitution profile, with nicotine failing to substitute for sucrose, but sucrose substituting for nicotine. However, this asymmetry may be due to the concurrent availability of two types of food reinforcers that economically interacted with each other. That is, the substitutability of chow for sucrose may have overshadowed the ability of nicotine to substitute for sucrose. Thus, the nature and degree to which drugs serve as a behavioral economic substitute for a food reinforcer might depend upon the the number and types of food reinforcers that are concurrently available and the behavioral economic relationship between them.

The use of the Cross-price Demand model and CRI_n measure to quantify the degree and nature of the relationship among reinforcer commodities provided several advantages over a linear regression (Johnson & Bickel, 2003) or purely visual analysis (e.g., (Wade-Galuska et al., 2007)), which are the more traditional approaches for assessing the economic relationship between reinforcers. A primary advantage of the Cross-price Demand model over linear regression and visual analysis, is that it allows for a more precise quantification of the nature of the economic relationship between reinforcers, both in terms of price sensitivity (β; steepness of the slope) and the absolute magnitude of the relationship (CRI value). Another advantage of the Cross-price model is that it provides a more accurate fit to the initial and final consumption, even when consumption plateaus. For example, the derived initial and final consumption levels from a linear regression fit becomes more inaccurate the longer consumption is measured beyond the point where a plateau is reached.

The cross-price demand model also has limitations that need to be addressed either through the parameter constraints used to fit the model or by modifications to the model itself. For instance, the present study limited β to values between 0 and 1 to prevent sharp upward or downward curve fits at low unit prices, which occurred with flat consumption data (i.e., an independent commodity) because it produced slightly higher R² values. The main issue with such fits is that it causes Q_{0 cross} (the y-intercept) to become out-of-line with the data, especially among linear horizontal fits. Another issue related to fitting flat consumption data is that the R² values become quite low (e.g., between 0 and .2), which indicates that the variance explained by the parameters of the cross-price demand equation do not contribute much more than that of a horizontal line through the mean of the data points (Kvalseth, n.d.). Thus, while low R² values may be the result of a poor fit, it may also simply indicate that the alternative reinforcer functioned as an independent to the primary commodity. Lastly, it is essential that consumption of the alternative commodity becomes stable at high unit prices of the primary commodity to provide an asymptote for the model to fit properly (i.e., properly estimate Q_alone), as too few data points will result in poor Q_alone estimates (i.e., extremely high or low values).

There are a number of limitations of the present study. First, only male rats were used, leaving open the question of whether there are sex differences in substitution between nicotine and food reinforcers. Second, the sample size was somewhat low, which may have limited the statistical power to detect certain effects (e.g. percent increase in sucrose consumption vs chow consumption during nicotine demand assessment). Third, the lever associated with each reinforcer was not counter balanced across subjects, so any position biases may have confounded the results. Fourth, rats were socially isolated. Because social factors can influence drug self-administration, this may have affected the results. Fifth, within-session temporal patterns of drug and food intake were not analyzed, which was beyond our initial interest in daily patterns of intake. Given that there is a clear circadian rhythm to nicotine self-administration in rats (e.g., Harris et al., 2007; O’Dell et al., 2007), within-session analysis could reveal stronger or weaker interactions between nicotine and food reinforcers at different times of the day. Sixth, flavors and sensory factors that may have differed between chow and sucrose pellets were not controlled and my have played role in the differential substitution of sucrose and chow for nicotine. For example, differential substitution may be lost if chow and sucrose pellets were the same flavor (e.g. chocolate). Seventh, the relatively short duration of nicotine demand assessment does not model the longer-term nicotine deprivation that would occur during smoking cessation. Thus, the present findings may not generalize to smokers making a quit attempt. However, the purpose of the present study was to examine behavioral economic interactions, not to model smoking cessation per se.

Despite these limitations, the present study demonstrates an important behavioral economic interaction (i.e., asymmetric substitution) between nicotine and sucrose that has important clinical and public health implications. Some smokers exhibit a large (15 kg) increase in body weight during smoking cessation that could begin to limit or delay the benefits of quitting smoking by increasing the risk for obesity-related diseases (Laguna et al., 2014; Siervo et al., 2014). Because the beneficial effect of quitting smoking has been shown to decrease proportionally for every kilogram of weight gained (Chinn et al., 2005; Hjellvik et al., 2010), it may take several years for this this subpopulation of ex-smokers to decrease their risk for diabetes (Bolego et al., 2002; Luo et al., 2012), coronary heart disease (Bolego et al., 2002), hypertension (Iso et al., 2005; Janzon et al., 2004), and impaired lung function (Willemse et al., 2004). The present findings support the notion that the marked weight gain in some ex-smokers may be due, in part, to the substitution of carbohydrates for decreases in nicotine intake. The present study provides a nonhuman model for investigating potential mechanisms underlying the substitutability of carbohydrates for nicotine/smoking and identifying interventions to reduce it, so that former smokers achieve the greatest health benefit from smoking cessation.

Highlights.

• The substitutability of sucrose for nicotine was examined in rats using behavioral economic analysis of concurrent sucrose pellet, chow pellet, and intravenous nicotine consumption.

• Sucrose served as a partial substitute for nicotine, but nicotine failed to substitute for sucrose.

• Sucrose was a stronger partial substitute for nicotine than chow in most rats.

• Findings demonstrate a behavioral economic mechanism that may mediate the selective increase in carbohydrate intake that can occur during smoking cessation.

Data availability

Data will be made available from the corresponding author upon reasonable request.