, Volume 169, Issue 1, pp 68–76

Operant responding for a visual reinforcer in rats is enhanced by noncontingent nicotine: implications for nicotine self-administration and reinforcement


  • Eric C. Donny
    • Department of Psychiatry, Behavioral Pharmacology Research UnitJohns Hopkins University School of Medicine
  • Nadia Chaudhri
    • Department of NeuroscienceUniversity of Pittsburgh
    • Department of PsychologyUniversity of Pittsburgh
  • F. Fay Evans-Martin
    • Department of PsychologyUniversity of Pittsburgh
  • Sheri Booth
    • Department of PsychologyUniversity of Pittsburgh
  • Maysa A. Gharib
    • Department of PsychologyUniversity of Pittsburgh
  • Laure A. Clements
    • Department of PsychologyUniversity of Pittsburgh
  • Alan F. Sved
    • Department of NeuroscienceUniversity of Pittsburgh
Original Investigation

DOI: 10.1007/s00213-003-1473-3

Cite this article as:
Donny, E.C., Chaudhri, N., Caggiula, A.R. et al. Psychopharmacology (2003) 169: 68. doi:10.1007/s00213-003-1473-3



Current conceptualizations of drug reinforcement assume that drug-taking behavior is a consequence of the contingent, temporal relationship between the behavior and drug reward. However, stimulant drugs also potentiate the rewarding effects of other reinforcers when administered noncontingently.


These studies were designed to determine whether noncontingent nicotine enhances the reinforcing properties of a nonpharmacological reinforcer and whether this direct effect facilitates operant behavior within the context of a nicotine self-administration procedure.


Rats self-administered nicotine or food, or received noncontingent nicotine, saline, or food either with or without a response-contingent, unconditioned reinforcing visual stimulus (VS).


Noncontingent nicotine, whether delivered as discrete injections based on a pattern of self-administered nicotine or as a continuous infusion, increased response rates maintained by the VS. There were no significant differences in responding by animals that received contingent compared with noncontingent nicotine when a VS was available. This increase was not observed in the absence of the VS or as a consequence of noncontingent food delivery. Operant behavior was equally attenuated and reinstated by the removal and subsequent replacement of contingent and noncontingent nicotine. Nicotine supported self-administration in the absence of response-contingent, nicotine-paired stimuli; however, response rates were drastically reduced compared with nicotine self-administration with the VS.


Nicotine influences operant behavior in two ways: by acting as a primary reinforcer when it is contingent upon behavior, and by directly potentiating the reinforcing properties of other stimuli through a nonassociative mechanism. Nicotine self-administration and smoking may be largely dependent upon this later action.




A basic tenet of behavioral research on addiction is that drug-taking behavior of both humans and animals is the result of a predictable temporal relationship between the behavior and drug reward. This dictum has been widely employed as the principle explanation for why people smoke tobacco—smoking results in the rapid delivery of nicotine to the brain, and the consequent neuropharmacological effects of nicotine reinforce smoking behavior (USDHHS 1988).

The hallmark test for drug reinforcement in laboratory animals is self-administration. The principle of contingency is a critical component of this test; response-contingent presentation of the drug should engender more robust operant behavior than response-independent drug delivery (Meisch and Lemaire 1993). The demonstration that drug-delivery must be contingent on the animal's behavior to support self-administration illustrates a central feature of instrumental behavior (Balleine and Dickinson 1998) and helps to eliminate alternative, nonassociative, explanations of drug-seeking behavior (e.g., non-specific locomotor activation; Meisch and Lemaire 1993). Nicotine, like other drugs of abuse, is self-administered by a variety of animal species (Corrigall and Coen 1989; Goldberg et al. 1981; Henningfield and Goldberg 1983b; Rose and Corrigall 1997). Nicotine self-administration is dose- and schedule-dependent (Corrigall and Coen 1989; Donny et al. 2000; Shoaib et al. 1997), extinguishes when nicotine is replaced with saline (Corrigall and Coen 1989; Shoaib et al. 1997), and, in the absence of other reinforcing stimuli, is dependent on nicotine being response-contingent (Donny et al. 1998). Models of nicotine self-administration are well-established and have been used to investigate the behavioral, environmental, and neurophysiological underpinnings of nicotine reinforcement (e.g., Caggiula et al. 2001; Corrigall et al. 1992; Picciotto et al. 1998).

However, other research suggests that some drugs can enhance responding for reinforcing stimuli by a mechanism that does not depend on a contingent relationship with either the stimuli or the behavior. For example, it is well-established that nicotine and other stimulants can directly increase low rates of schedule-controlled behavior (Byrd 1979; Dews 1958; Hendry and Rosecrans 1982). Furthermore, noncontingent administration of psychostimulants such as amphetamine, cocaine, and pipradrol can enhance responding for stimuli that have previously been associated with primary rewards (Beninger et al. 1981; Hill 1970; Robbins 1976, 1977; Robbins and Koob 1978; Robbins et al. 1983). Phillips and Fibiger (1990) reviewed the literature on the reward-enhancing properties of cocaine and concluded that these effects may provide an additional mechanism driving cocaine abuse, which is distinct from the primary reinforcing effects of the drug. With some notable exceptions (e.g., Whitelaw et al. 1996; Markou et al. 1999), relatively little attention has been paid to the possibility that the reward-enhancing effects of stimulants may contribute to stimulant self-administration.

The aim of the present studies was to determine if nicotine produced reward-enhancing effects like those observed for cocaine and amphetamine and to begin to evaluate whether such effects might contribute to operant responding within the context of an animal model of self-administration. We examined the rate of responding maintained by either self-administered or response-independent nicotine under conditions in which an unconditioned reinforcing visual stimulus (Caggiula et al. 2002) was either present or absent. The results demonstrated that noncontingent administration of nicotine greatly enhanced responding for reinforcing stimuli. These findings have important implications for both understanding the factors controlling nicotine-seeking behavior and interpreting standard laboratory evaluations of drug reinforcement.

Materials and methods


Male Sprague Dawley rats (Harlan Farms), 41–44 days old and weighing between 200 and 225 g upon arrival, were individually housed in a temperature-controlled environment on a 12-hr reversed light/dark cycle. Upon arrival, all animals were placed on an unrestricted diet during one week of habituation to the laboratory. After training (described below) and for the remainder of the study, all animals received 20 g of food per day. Unlimited access to water was available throughout all experiments. Animals were 60+ days old at the start of the experiments. Separate cohorts of animals were used for each of the experiments described below.


Lever training and all subsequent experimental sessions took place in a 25×31×28 cm operant conditioning chamber (BRS/LVE Model RTC-020) with identical inactive and active levers, a cue light located 5 cm above the active lever, an overhead chamber light, and a pellet trough. For experimental sessions, all animals were connected to a drug-delivery swivel system that allowed nearly unrestricted movement in the chamber. An interfaced computer software package (Med Associates, MED-PC IV) was used to record active lever responses, inactive lever responses, and reinforcements. A constant background noise of approximately 75 dB that was produced by exhaust fans located within each sound-attenuating chamber masked the auditory cues associated with food/drug delivery and ambient noise.


Following habituation to the colony room, rats were food deprived for 24 h and then trained to lever press on the right (active) lever for 45 mg food pellets. Training consisted of a single 20-min habituation session in the experimental chamber, a 25-min magazine training session, and a session that began with hand shaping, during which animals received approximately 20 pellets as a consequence of responding on the active lever, and ended with a programmed fixed ratio (FR) 1 with a maximum of 75 food reinforcements. Responding on the left (inactive) lever had no scheduled consequence. In all of the experiments described here, rats were trained in the absence of any scheduled changes in stimulus conditions (i.e., the cue light remained off and a red chamber light was illuminated to allow monitoring by the experimenters).


Following training all animals were anesthetized with halothane and implanted with jugular catheters. Animals were allowed at least 7 days to recover from surgery prior to the start of the experimental sessions. For the first 2 weeks after surgery, rats were treated with both heparin and streptokinase to help maintain catheter patency, and the antibiotic ticarcillan plus clavulanate to reduce postsurgical infections (see Donny et al. 1999 for details). Thereafter, catheters were flushed once daily with 0.1 ml sterile, heparinized saline (30 U/ml) on non-testing days (weekends), and both prior to (10 U/ml) and following (30 U/ml) each session on testing days.

Experiment 1: the effects of contingent and noncontingent (i.e., yoked) nicotine on operant responding in the presence or absence of a behaviorally-contingent visual stimulus

The purpose of experiment 1 was to (1) determine whether noncontingent nicotine administration potentiates responding maintained by an unconditioned, reinforcing, visual stimulus (VS), and (2) begin evaluating the potential influence of this effect on self-administration. Animals in experiment 1 were divided into six groups following food training. In one group (contingent-nicotine + VS), fulfillment of the schedule requirement resulted in the delivery of an intravenous infusion of 0.03 mg/kg nicotine bitartate (all infusions were delivered in a volume of 0.1 ml/kg over approximately 1 s; dose reported as free base weight) and a VS paired with each infusion. A schematic of the relationship between operant behavior, infusions, and changes in the VS is shown in Fig. 1. In the contingent-nicotine + VS condition, nicotine and the VS were always presented together and dependent upon the animal's behavior (see Fig. 1). The VS, used previously in studies of nicotine self-administration (Donny et al. 1998, 2000), consisted of the onset of a white cue light for 1 s and the concurrent offset of a white chamber light for 1 min. A second group (contingent-nicotine + no VS) also self-administered 0.03 mg/kg nicotine, but without any scheduled changes in stimulus conditions. The third and fourth groups were allowed to self-administer saline either paired with the VS (contingent-saline + VS) or without the VS (contingent-saline + no VS). Animals in the fifth group (noncontingent-nicotine + VS) were paired with individuals in the contingent-nicotine + VS group and received the same number of nicotine infusions at identical times during each session (i.e. yoked). Figure 1 illustrates the differences between this condition and the contingent-nicotine + VS group. Nicotine infusions in this group were contingent upon the partner's responding and not upon their own lever pressing. However, lever pressing by animals in this group resulted in the response-contingent presentation of the VS. A sixth group also received yoked nicotine infusions (matched to the same individual animals in the contingent-nicotine + VS condition), but lever pressing had no scheduled consequence (noncontingent-nicotine + no VS). The schedule of reinforcement for nicotine/saline and VS presentations was increased sequentially from an FR 1 (days 1–5), through an FR 2 (days 6–13) to an FR 5 (days 14–23). All response-contingent reinforcements (i.e., contingent nicotine, saline and/or VS) were followed by a 1-min time out period during which responding had no consequence. Noncontingent presentations of nicotine or saline were not followed by a time out period. All experimental sessions lasted 1 h.
Fig. 1.

Operant responding on the active lever, infusions, and stimulus changes under the contingent-nicotine + VS and noncontingent-nicotine + VS conditions on an FR 5 schedule

Experiment 2: replication of the effects of noncontingent (i.e., yoked) nicotine and comparison with noncontingent saline and food

The purpose of experiment 2 was to (1) further evaluate the tendency for noncontingent nicotine to elevate responding for the VS by comparing these effects with noncontingent saline control conditions, (2) reexamine the similarities and/or differences between contingent and noncontingent nicotine in the presence of the VS, and (3) determine whether noncontingent food would produce a similar elevation in response rates. Animals in experiment 2 were divided into seven groups following food training. Two of these groups were identical to those described in experiment 1 (contingent-nicotine + VS and noncontingent-nicotine + VS). Two additional groups received noncontingent infusions of saline that were yoked to the contingent-nicotine + VS group while their responding resulted in either the contingent presentation of the VS (noncontingent-saline + VS) or no consequence (noncontingent-saline + no VS). A fifth group of animals lever pressed for 45-mg food pellets paired with the VS (contingent-food + VS). Individuals in the sixth group (noncontingent-food + VS) received yoked food pellets (i.e., controlled by animals in the contingent-food + VS condition) while responding for contingent presentations of the VS. The seventh group (noncontingent-food + no VS) also received yoked food pellets, but lever pressing in this group had no consequence. The schedule of reinforcement for contingent nicotine, food, and VS presentations was an FR 1 for days 1–5, an FR 2 for days 6–13, and an FR 5 for days 14–20. A 1-min time out period followed all response-contingent reinforcers. All experimental sessions lasted 1 h.

Experiment 3: changes in responding after the removal and replacement of self-administered versus yoked nicotine

The purpose of experiment 3 was to (1) further evaluate the direct effects of noncontingent nicotine on cue-maintained responding by substituting saline for nicotine in a within-subjects design, and (2) compare these effects with extinction and reacquisition in animals self-administering nicotine. Animals were divided into the two groups described in experiment 1 (contingent-nicotine + VS and noncontingent-nicotine + VS). Following a 20-day acquisition period identical to experiment 1, saline was substituted for nicotine for 3 days, and then nicotine was reinstated from days 24–28 in both groups. The response-contingent presentation of the VS remained available throughout the experiment. The schedule of reinforcement for nicotine/saline and for the VS was an FR 5 throughout the maintenance, extinction, and reacquisition phases. A 1-min time out period followed all response-contingent reinforcers. All experimental sessions lasted 1 h.

Experiment 4: the effects of continuously infused nicotine on cue-maintained responding

The purpose of experiment 4 was to (1) determine whether a continuous intravenous infusion of nicotine would potentiate responding maintained by the VS in a manner similar to pulsed infusions of nicotine, and (2) evaluate changes in cue-maintained responding during saline substitution. Animals were divided into four groups. Three of the groups were identical to the groups reported in experiment 1 and 3, including a contingent-nicotine + VS group, a noncontingent-nicotine + VS group, and a noncontingent-saline + VS group. The fourth group received a noncontingent, continuous infusion of nicotine while responding was reinforced by the VS (continuous-nicotine + VS). Due to experimenter error, the cumulative dose of continuous nicotine administered each session was approximately one third the dose self-administered by animals given access to 0.03 mg/kg/inf. Therefore, direct comparisons of either the contingent-nicotine + VS or noncontingent-nicotine + VS and the continuous-nicotine + VS conditions should be made with caution since the conditions differ across multiple parameters (e.g., methods of nicotine delivery, dose). The total nicotine delivery per 1-hr session was 0.039 mg/kg for days 1–2, 0.19 mg/kg for days 3–5, and 0.23 mg/kg for days 6–29. The concentration of nicotine base dissolved in physiological saline was 0.03, 0.1, and 0.1 mg/ml and the flow rate was 1.3, 1.9, and 2.3 ml/h for days 1–2, 3–5, and 6–29, respectively. Following a 20-day acquisition period, saline was substituted for nicotine for 6 days, and then nicotine was replaced from days 27–29 in all three nicotine groups. The contingent-saline + VS group was not run after day 26. The response-contingent presentation of the VS remained available throughout the experiment. The schedule of reinforcement for contingent nicotine/saline and/or VS presentations was increased sequentially from an FR 1 (days 1–5), through an FR 2 (days 6–13) to an FR 5 (days 14–29). A 1-min time out period followed all response-contingent reinforcers. All experimental sessions lasted 1 h.

Experiment 5: a dose-effect analysis of nicotine self-administration without cues

The purpose of experiment 5 was to determine whether response-contingent nicotine maintained operant behavior in the absence of other stimuli. Four groups of animals acquired nicotine self-administration without any programmed visual or auditory stimuli under the following schedule of reinforcement: FR 1 (days 1–5), FR 2 (days 6–8), FR 5 (days 9–20). A 1-min time out period followed all nicotine injections. Each group was assigned a different dose of nicotine (0.015, 0.03, 0.06, or 0.09 mg/kg/infusion) that remained constant throughout the study. Otherwise all conditions were identical to those described for contingent-nicotine + no VS above. All experimental sessions lasted 1 h.

Statistical analyses

Statistical analyses were conducted using the mean of the last 2–3 days of each schedule of reinforcement (FR 1, FR 2, FR 5) or test phase (maintenance, extinction, or reacquisition). This approach allowed for comparison across conditions with an unequal number of sessions and focused on stable behavior. It is important to point out that this approach does not capture the dynamic changes in behavior that occur over time within each schedule of reinforcement or test phase (e.g., the rate of extinction). The 3-day mean was used in all experiments except experiments 3 and 5; in these cases, a 2-day mean was used to avoid using the 1st day of a 3-day extinction period (experiment 3) or the 1st day of 3 days on an FR 2 (experiment 5).

Data from all experiments were analyzed using ANOVA with either schedule of reinforcement (experiments 1, 2, 4, and 5) or extinction phase (experiment 3 and 4) as the within-subjects factor and group (experiments 1–4) or dose (experiment 5) as the between-subjects factor. Analysis of extinction data from experiment 4 did not include data from contingent-saline + VS. Pre-planned comparisons between groups utilized targeted two-factor ANOVAs (schedule/phase and group/dose) with the between-subject factors confined to the two conditions of interest, followed by paired and independent sample t-tests. The α level was set to 0.05.


Experiment 1

Lever pressing by both the contingent-nicotine + VS and the noncontingent-nicotine + VS groups rose rapidly across the 20-day acquisition period, reaching robust, stable rates that were greater than those maintained by nicotine or the VS alone (Fig. 2). ANOVA indicated that active response rates significantly differed by group [F(5,43)=23.54, P<0.001], schedule [F(2,86)=28.09, P<0.001] and the group-by-schedule interaction [F(10,86)=8.88, P<0.001]. Response-contingent presentation of the VS increased active response rates compared with the no VS conditions both in the presence of contingent saline (P<0.01) and noncontingent nicotine (P<0.001), indicating that the VS was functioning as a reinforcer. Noncontingent nicotine tended to facilitate responding for the VS compared with contingent-saline + VS, although this difference only reached statistical significance on an FR 2 (P<0.05). The facilitating effects of noncontingent nicotine on response rates for the VS were significantly different from those of contingent nicotine on an FR 1 (P<0.05); however the groups converged by the end of FR 2 and did not significantly differ throughout the remainder of the study. Active response rates in the noncontingent-nicotine + no VS were low throughout the experiment and did not differ from either the contingent-saline + no VS or the noncontingent-saline + no VS from experiment 2 (see below). Similarly, at this dose (see experiment 5 for full dose-response analyses), contingent nicotine in the absence of the VS failed to significantly increase active lever response rates over contingent-saline + no VS. The mean (± SEM) number of nicotine infusions in the contingent-nicotine + VS (and the yoked, noncontingent-nicotine groups) was 23.0 (±1.68), 27.4 (±1.25), and 24.3 (±1.81) for the last 3 days of FR 1, FR 2 and FR 5, respectively. In contrast, the mean (± SEM) number of infusions by contingent-nicotine + no VS was 6.2 (±0.85), 6.5 (±1.38), and 2.4 (±0.63) and similar to the rate of infusions maintained in the contingent-saline + no VS condition. Mean (± SEM) response rates on the inactive lever during FR 5 were 33.8 (±3.6), 16.7 (±2.3), 7.3 (±1.3), 9.7 (±1.8), 1.5 (±0.4), and 5.0 (±0.6) for the contingent-nicotine + VS, noncontingent-nicotine + VS, noncontingent-nicotine + no VS, contingent-nicotine + no VS, contingent-saline + VS, and contingent-saline + no VS, respectively.
Fig. 2.

Effects of nicotine and visual stimuli (VS) on active lever responding (including responding during the time out period). Results are mean ± SEM of data obtained from 7–10 animals per group. Schedule of reinforcement is indicated below the abscissa

Experiment 2

Experiment 2 replicated the finding that responding maintained by contingent and noncontingent nicotine was similar in the presence of a VS. This study also clearly demonstrated that responding for the VS was higher when accompanied by noncontingent nicotine compared with noncontingent saline; in contrast, noncontingent food produced rates of behavior that were similar to noncontingent saline. The nicotine and food data are presented in separate panels (Fig. 3) to accommodate the large differences in the maximal rates of behavior. ANOVA revealed that active response rates significantly differed by group [F(6,55)=73.72, P<0.001], schedule [F(2,110)=61.49, P<0.001] and the group-by-schedule interaction [F(12,110)=20.53, P<0.001]. As in experiment 1, the VS functioned as a reinforcer, potentiating responding in the presence of both noncontingent saline (P<0.05) and noncontingent food (P<0.01). Active response rates in the presence of the VS were not statistically different in the contingent or noncontingent-nicotine conditions and were significantly potentiated compared with noncontingent-saline + VS (P<0.005 for overall group effect, FR 2, and FR 5). In contrast to the effects of nicotine, rats receiving response-contingent food pellets paired with the VS responded at rates (Fig. 3, lower panel) that were substantially higher than noncontingent-food + VS (P<0.001 for overall group effect, group-by-schedule interaction, and all FR schedules). The delivery of noncontingent-food pellets in the presence of the response-contingent VS yielded response rates that were not significantly different from the rate of responding in the noncontingent-saline + VS condition. Mean (± SEM) response rates on the inactive lever during FR 5 were 20.9 (±3.5), 14.1 (±2.1), 7.4 (±1.9), 5.2 (±1.5), 5.7 (±1.5), 4.2 (±0.8), and 4.7 (±0.5) for contingent-nicotine + VS, noncontingent-nicotine + VS, noncontingent-saline + VS, noncontingent-saline + no VS, contingent-food + VS, noncontingent-food + VS, and noncontingent-food + no VS, respectively.
Fig. 3.

Effects of nicotine, food, and visual stimuli (VS) on active lever responding (including responding during the time out period). Results are mean ± SEM of data obtained from 8–10 animals per group. Upper panel illustrates the effects of nicotine and/or saline. Lower panel illustrates the effects of food. Schedule of reinforcement is indicated below the abscissa

Experiment 3

Responding for the VS was similarly attenuated in the contingent-nicotine + VS and noncontingent-nicotine + VS conditions when saline was substituted for nicotine and reinstated when nicotine was reintroduced (Fig. 4). ANOVA on active lever responding revealed a significant effect of phase [F(1,15)=78.15, P<0.001], but no effect of group or the group-by-phase interaction (P>0.05). In both conditions, the rates of responding on the reinforced lever decreased significantly when nicotine was substituted with saline (P<0.001) and increased after nicotine was reintroduced (P<0.001); there were no significant differences between the groups. Responding on the inactive lever was similar and significantly lower than active responding for both groups (P<0.001).
Fig. 4.

Effects of removing and subsequently replacing nicotine on active lever responding (including responding during the time out period). Results are mean ± SEM of data obtained from 8–9 animals per group

Experiment 4

The continuous infusion of a relatively low dose of nicotine resulted in a potentiation of responding maintained by the VS that was similar to the effect of larger doses of both response-contingent and yoked nicotine (Fig. 5). ANOVA on active responses during acquisition revealed a significant effect of group [F(3,27)=11.72, P<0.001], schedule [F(2,54)=189.21, P<0.001], and the group-by-schedule interaction [F(6,54)=11.58, P<0.001]. Pairwise comparisons of acquisition data revealed that continuous-nicotine + VS increased response rates compared with contingent-saline + VS (P<0.001 overall and at each FR schedule). As observed in the previous experiments, the contingent-nicotine + VS and noncontingent-nicotine + VS failed to significantly differ from each other, and both displayed a higher overall rate of responding (P<0.05) and potentiated responding during individual schedules of reinforcement (FR 1, contingent-nicotine + VS only: P<0.05; FR 2: P<0.05; FR 5: P<0.001) compared with contingent-saline + VS. The similarities between contingent-nicotine + VS and noncontingent-nicotine + VS remained when time-in and time-out active responses were considered separately. The percentage of active responses that occurred during the time out period decreased over the acquisition period in all three nicotine conditions (contingent, noncontingent and continuous nicotine), representing approximately 19–26% of the total number of responses made during the last 3 days on an FR 5. Inactive responding in the contingent-nicotine + VS and contingent-saline + VS was low (7.6±1.0 and 3.9±0.1 on an FR 5, respectively). Noncontingent-nicotine + VS and continuous-nicotine + VS displayed elevated rates of inactive responding (37.2±12.6 and 48.5±12.4 on an FR 5, respectively) as a result of two animals in each condition responding at unusually high rates (>100 responses). The remaining animals displayed relatively low rates of inactive responses.
Fig. 5.

Effects of response-contingent, yoked, or continuous nicotine or contingent saline on active lever responding (including responding during the time out period) for VS. Acquisition (days 1–20) was followed by 6 days in which saline was substituted for nicotine (days 21–26) and then 3 days in which nicotine was replaced (days 27–29). Results are mean ± SEM of data obtained from 7–9 animals per group

Analysis of extinction confirmed that it was the continuous infusion of nicotine that increased response rates in animals responding for the VS (Fig. 5). ANOVA of maintenance, extinction, and reacquisition revealed a significant effect of phase [F(2,38)=97.25, P<0.001], but no effect of group [F(2,19)=.064, NS] or the group-by-phase interaction [F(4,38)=.722, NS]. Substituting saline for nicotine decreased the rate of responding in the contingent-nicotine + VS, noncontingent-nicotine + VS, and continuous-nicotine + VS groups (P<0.001 compared with maintenance) and replacing nicotine increased the rate of operant responding in all three nicotine groups (P<0.005 compared with extinction). Inspection of data from the contingent-saline + VS condition revealed continued stable rates of active lever responding from day 14 through day 26.

Experiment 5

Nicotine self-administration in the absence of cues was dose-dependent, producing an inverted-U-shaped curve that peaked at 0.06 mg/kg/infusion (Fig. 6). ANOVA of active responses revealed the following main effects and interactions: dose [F(3,26)=4.37, P<0.05], schedule [F(2,52)=19.54, P<0.001], dose by schedule [F(6,52)=2.64, P<0.05]. The lowest dose, 0.015 mg/kg/infusion, maintained relatively low rates of active responding that were not significantly different from inactive lever response rates. The two middle doses (0.03 and 0.06 mg/kg/infusion) supported more active than inactive lever responding on all schedules of reinforcement (P<0.05). The largest dose, 0.09 mg/kg/infusion, resulted in active response rates that were only significantly greater than inactive responses on an FR 5 (P<0.01). Pairwise comparisons between doses revealed no differences between 0.015 and 0.03 mg/kg/infusion. There were significantly more active responses in the 0.06 mg/kg/infusion group compared with 0.015 mg/kg/infusion across schedules of reinforcement (P<0.001) as well as for FR 2 and FR 5 when analyzed separately (P<0.05). Comparison of the largest dose, 0.09 mg/kg/infusion, to 0.015 mg/kg/infusion indicated significant group differences that varied across schedule of reinforcement (i.e., group-by-schedule interaction; P<0.05), reaching significance on an FR 2 (P<0.05).
Fig. 6.

Self-administration of different doses of nicotine without additional programmed stimuli. Results are mean ± SEM of data obtained from 6–8 animals per group. Adjacent symbols represent the mean active lever response rates (including responding during the time out period) on the last 2 days on an FR1, FR2, and FR5 schedule of reinforcement, respectively. Bars represent infusions obtained during the last 2 days on an FR5


These findings support the hypothesis that nicotine enhances the reinforcing properties of other stimuli. This action of nicotine was demonstrated by the nicotine-induced increase in responding for a concurrently available, reinforcing, VS. The increase in responding was dependent on the availability of the VS, did not occur with noncontingent-food delivery, and was under the control of nicotine delivery as demonstrated by saline substitution. The critical observation made here was that operant responding was maintained at high levels by nicotine that was neither temporally nor causally associated with behavior, indicating that this effect is distinct from the actions of nicotine as a primary reinforcer (Phillips and Fibiger 1990). This finding does not contradict the hypothesis that nicotine is the "primary psychoactive ingredient driving smoking" (USDHHS 1988), but rather suggests that nicotine may support behavior in two ways: by acting as a primary reinforcer and by directly potentiating the reinforcing effects of other stimuli.

Previous research has shown that psychostimulants such as amphetamine, cocaine, and pipradrol (Hill 1970; Phillips and Fibiger 1990; Robbins and Koob 1978; Stein 1964) can enhance the reinforcing effects of other stimuli through nonassociative mechanisms. The present results suggest that nicotine may also have reinforcement-enhancing effects that may contribute to nicotine's control over behavior. This effect was observed both when nicotine administration was yoked to mimic the dose and pattern of self-administered nicotine and when a relatively low dose of nicotine (e.g., approximately one third the cumulative self-administered dose at 0.03 mg/kg/infusion) was slowly infused throughout the experimental session. The effectiveness of both yoked pulsed and continuous infusions of nicotine strongly supports the notion that the reinforcement enhancing effects of nicotine were nonassociative in nature and not the result of intermittent, chance associations between nicotine delivery and either operant behavior or presentation of the VS. Furthermore, although the comparison to food reinforcement is limited by the fact that food requires consumatory behavior, the observation that noncontingent-food delivery failed to alter responding suggests that the increase in response rates was a direct, pharmacological action of nicotine and not a property of all reinforcers.

Although the present study examined an unconditioned reinforcing light stimulus (see Caggiula et al. 2002 for detailed discussion of this stimulus condition), the reinforcement-enhancing effects of nicotine are likely to extend to conditioned reinforcers. Indeed, most evidence that psychomotor stimulants enhance the effectiveness of other reinforcers has focused on conditioned reinforcement (e.g., Robbins 1976; Robbins and Koob 1978; Taylor and Horger 1999). A similar effect of nicotine would be important since nicotine-related stimuli are hypothesized to play a critical role in both nicotine self-administration in animals and smoking in humans (Caggiula et al. 2001; Rose and Levin 1991). Nicotine self-administration in animals and smoking behavior in humans is assumed to result from the combined primary reinforcing effects of nicotine and the secondary reinforcing effects of nicotine-related stimuli. However, the present data suggest a critical interaction. The strength of the sensory stimuli as reinforcers may be greatly potentiated by nicotine, not only because nicotine has been repeatedly paired with these cues, but because nicotine acts directly to potentiate the reward value of those stimuli. Phillips and Fibiger (1990) recognized a similar effect in their review of the reward-enhancing properties of cocaine. After noting that the "conditioned stimuli" used in these studies often exhibit little or no reinforcing effects when presented alone, they pointed out that "the effects of conditioning are quite evident under the influence of the drug." They proposed that "under certain conditions there are latent conditioned rewarding effects that are only revealed after administration of a psychomotor stimulant" (see p. 275 in Phillips and Fibiger 1990). Likewise, the influence of nicotine-related stimuli may be greatest in the presence of the reinforcement-enhancing effects of nicotine.

Whether nicotine produces lasting changes in the reinforcing effects of other stimuli is unclear. Research has demonstrated that smoking cues continue to elicit positive subjective effects and reduce craving and withdrawal in the absence of nicotine (e.g., Pickworth et al. 1999; Rose et al. 2000). The generally accepted explanation for these effects is that smoking stimuli have become conditioned reinforcers as a consequence of repeated pairings with nicotine. However, these stimuli have always been experienced in the presence of nicotine, making it impossible to disentangle the simple conditioned reinforcing effects from a history of experiencing those effects in the presence of nicotine. Indirect evidence presented here suggests that some carry-over effects of nicotine may be present. A history of responding for the VS in the presence of nicotine (i.e., noncontingent-nicotine + VS), yielded response rates without nicotine (i.e., 1st day of extinction) that were more than 40% greater than those maintained by noncontingent-saline + VS (i.e., when animals had never experienced the VS in conjunction with nicotine). Although additional studies are required to rule out alternative explanations (e.g., behavioral momentum), these data suggest that reinforcing stimuli that are consistently experienced in the presence of nicotine may develop a greater reinforcing value than would be reached without this history.

These findings have important implications for models of drug self-administration. Most models of drug self-administration employ drug-paired, visual and/or auditory stimuli that would be expected to either have primary reinforcing effects (Lowe and Williams 1969; Stewart and Hurwitz 1958; Stewart 1960) or acquire conditioned reinforcing properties (e.g., Everitt et al. 2001). The reinforcement-enhancing effects of a drug may facilitate operant behavior within the context of the self-administration model in a manner that is not dependent on the primary reinforcing properties of the drug per se (Phillips and Fibiger 1990). If this is true, changes in nicotine and other drug self-administration that occur as a consequence of neurophysiological, pharmacological, and behavioral manipulations may be attributable to changes in the reinforcement-enhancing effects of the drug and not necessarily to its primary reinforcing effects.

Data presented here, as well as previous research (Caggiula et al. 2002), demonstrate that nicotine supports self-administration in the absence of other reinforcers; however, there are large differences in self-administration behavior with and without response-contingent stimuli (Caggiula et al. 2001). A moderate dose of 0.03 mg/kg/infusion nicotine functions effectively as a robust reinforcer in the presence of the VS (e.g., Donny et al. 1998, 2000), but is only marginally reinforcing in its absence (Caggiula et al. 2002; data presented here). In experiment 5, a full analysis of nicotine self-administration without cues revealed response rates that peaked at a dose (0.06 mg/kg/infusion) that was three times larger than the peak dose in the presence of the VS (Donny et al. 2000). Furthermore, the maximal rate of nicotine self-administration is two to three times greater with the VS (Donny et al. 2000) than without it (experiment 5). Additional research from our laboratory has confirmed that the dose-response function for nicotine self-administration without cues is shifted sharply downward and to the right compared with when nicotine is paired with the VS (unpublished observations). These observations are consistent with reports in humans that puff-sized doses of intravenous nicotine (i.e., without smoking cues) produce only small increases in satisfaction and liking (Rose et al. 2000; Westman et al. 1996), but that larger doses produce moderate increases in positive subjective effects (Garrett and Griffiths 2001) and self-administration (Henningfield and Goldberg 1983a).

The degree to which nicotine-paired stimuli potentiate nicotine self-administration is likely to be related to the unconditioned reinforcing value of those stimuli and the ability of nicotine to potentiate their effects. A recent study by Caggiula et al. (2002) examined the acquisition of nicotine self-administration behavior under a variety of cue conditions. Cues that were relatively neutral (i.e., did not support operant behavior when tested alone) produced a small increase in self-administration, presumably because of their repeated association with nicotine. In contrast, unconditioned reinforcing stimuli produced a large increase in responding. These findings support the notion that, under certain stimulus conditions, a substantial portion of self-administration may be determined by a direct effect of nicotine on behavior reinforced by other stimuli.

Other evidence supports the notion that nicotine potentiates the reinforcing properties of other reinforcers. In animals, nicotine increases motivation to obtain food (Popke et al. 2000), potentiates alcohol and cocaine self-administration (Bechtholt and Mark 2002; Clark et al. 2001; Potthoff et al. 1983), and lowers the threshold for brain reward stimulation (Bauco and Wise 1994). Likewise, clinical studies have found that smoking often occurs in conjunction with other reinforced behavior (e.g., drinking alcohol; Bien and Burge 1990). Although these effects are often interpreted as being pharmacologically specific (e.g., nicotine-alcohol interactions), an alternative interpretation is that nicotine acts more broadly, potentiating the rewarding effects of reinforcing stimuli. Recent neurophysiological evidence is also consistent with a more general effect; the net GABAergic and glutamatergic influence on brain dopamine systems may shift towards a more excitable state following nicotine exposure (Mansvelder et al. 2002).

In conclusion, these studies support the proposal of an alternate action of nicotine that may operate in conjunction with its primary reinforcing effects to drive smoking behavior. Our data indicate that nicotine enhances the reinforcing value of other, nonpharmacological stimuli in a manner that is not dependent on a close temporal association between nicotine and either the stimuli, or the behavior controlling their delivery. The demonstration that nicotine produces both primary reinforcing effects and potent enhancement of the reinforcing effects of other stimuli suggests that basic research and treatment strategies predicated on nicotine acting principally as a primary reinforcer may be deficient.

Copyright information

© Springer-Verlag 2003