Cocaine reinstates extinguished food responding in male cynomolgus monkeys with a history of self-administering cocaine under a concurrent drug versus food choice paradigm

Abstract

The rate-dependency principle postulates that “stimulants” tend to increase low baseline rates of response and decrease high baseline responses. One exception to this principle is that stimulants do not typically increase responses that has been extinguished. However, drug self-administration studies showed that stimulant pretreatments reinstated extinguished drug-maintained responses. The goal of this study was to determine the effects of noncontingent cocaine on extinguished food-maintained responding in 3 male cynomolgus monkeys with extensive cocaine self-administration experience. Monkeys had a >4-year history of self-administering cocaine under a concurrent cocaine versus food schedule of reinforcement. For this study, only the discriminative stimulus (S^D) signaling a fixed-ratio 30 schedule of food reinforcement was studied, and the previous cocaine-associated S^D was not illuminated (inactive). In experiment 1, the effect of noncontingent food (1–5 pellets) and cocaine (0.03–0.3 mg/kg i.v.) on extinguished food-maintained responses was studied. In all 3 monkeys, 5 pellets and at least 1 cocaine dose significantly increased extinguished responding; the effects of cocaine were larger than those of food. In experiment 2, the behavioral mechanisms mediating these cocaine-induced increases in extinguished responses were studied. Following administration of 0.3 mg/kg cocaine, with no S^D illuminated, responses occurred on both the previously active (food) and inactive (cocaine) manipulanda. When cocaine was given and both S^Ds were illuminated, more responses occurred on the inactive (cocaine) versus previously active (food) switch. These findings suggest that a cocaine self-administration history can influence the direct behavioral effects of cocaine. Increases in previously extinguished nondrug-reinforced behaviors could have clinical implications related to relapse.

Significance Statement

Following long-term cocaine self-administration in the context of a nondrug alternative, noncontingent cocaine administration reinstated previously extinguished responses maintained by delivery of food pellets, suggesting that reinstatement studies may not be measuring relapse of “drug-seeking.” Importantly, the effects of self-administered cocaine on the brain and behavior cannot be determined in the absence of systematically studying ongoing behavior.

1. Introduction

The 1955 study by Peter Dews, showing that the behavioral effects of pentobarbital were different depending on the ongoing response rates (Dews, 1955), is considered one of the most important studies associated with the origin of the discipline of behavioral pharmacology (Branch, 1991). In that study, pentobarbital, a drug considered to be a central nervous system (CNS) depressant, increased high-rate, fixed-ratio (FR) responses by pigeons, while the same doses decreased low-rate, fixed-interval responses. The significance of this early finding was that a drug considered to be a CNS depressant could increase behavior, and the same doses of a drug can produce vastly different effects on responses, depending on the baseline (BL) rate of behavior. Since that seminal study, a great deal of research has focused on the rate-dependent effects of drugs (Kelleher and Morse, 1968). For example, drugs considered CNS stimulants, like cocaine and d-amphetamine, have been shown to decrease high rates of behavior at doses that increase low rates of responding (Dews and Wenger, 1977).

As it relates to “stimulant-like” drugs, exceptions to the ability of these drugs to increase low rates of responding have been noted (Kelleher and Morse, 1968). These exceptions are: (1) operant level, or the behavior measured before an operant response has been established and when there were no contingencies scheduled for any particular response; (2) punished response; and (3) extinguished operant response. As it relates to punished response, McKearney and Barrett (1975) showed that when punished food-maintained responding was studied under multiple schedules with a shock avoidance contingency in the other component, d-amphetamine would increase the punished response. Barrett (1977) followed that study by showing that simply a history of shock avoidance would alter the effects of d-amphetamine on punished response. These seminal studies showed that context and history can alter the effects of d-amphetamine on punished responding.

Because it relates to drugs like cocaine and d-amphetamine, increasing extinguished response and reinstating previous drug self-administration appears to be an exception to the rule that these drugs do not increase extinguished behaviors. One hypothesis for this reinstatement effect is that the drug that increases saline self-administration is simply substituting for the previously reinforcing drug in a manner related to drug discrimination. Banks et al (2007) tested the hypothesis that cocaine would increase extinguished food-maintained responding in a group of monkeys after a chronic history of self-administering cocaine (>2 years) and in a group of monkeys after a brief history of cocaine self-administration (0.27–0.58 mg/kg cumulative cocaine intakes). In that study, they first showed in the monkeys with a chronic history of self-administering cocaine under a concurrent schedule with food as the alternative that, when saline was substituted for cocaine, monkeys would choose food almost exclusively, and cocaine pretreatment dose-dependently increased the “cocaine” (actually saline) choice. In another group of cocaine-naïve monkeys, they showed that noncontingent cocaine would not increase extinguished food-maintained responding. They then trained those monkeys to self-administer cocaine, and after a brief history in which cocaine functioned as a reinforcer, they redetermined the noncontingent cocaine dose-response curve on extinguished food-maintained response: cocaine still did not increase response. These findings suggest that cocaine functioning as a reinforcer was not enough to alter the effects of noncontingent cocaine on extinguished food-reinforced response.

In their review of rate dependency, Dews and Wenger (1977) stated, “[T]here is more to behavior, even under a schedule of reinforcement maintaining much responding than average rate of responding; responses occur with changing frequency in real time, there are force and duration variables, there are stimulus control functions involving sensory components, and so on. Preoccupation with rate should not lead to the overlooking of other variables” (p. 173; emphasis added). Thompson (1984) and Thompson and Schuster (1968) introduced the principle of the behavioral mechanism of drug action. This refers to a description of a drug’s effect on a given behavioral system in terms of a general set of environmental principles regulating behavior. According to this principle, a drug effect can be explained when the investigators show that the effect was due to the action of the drug on 1 or more of the basic mechanisms that normally control behavior. These were described as stimulus, response, antecedent, and consequent variables (see Thompson, 1984).

A major goal of the present study was to further evaluate the ability of noncontingent cocaine to increase extinguished food-reinforced response. A limitation of the Banks et al (2007) study was that they did not test their hypothesis that noncontingent cocaine would increase extinguished food-reinforced response in monkeys with an extensive cocaine self-administration history. In the present study, monkeys with long histories of cocaine self-administration were studied. These monkeys had self-administered cocaine under several schedules of reinforcement, but before the present study began, they had been self-administering cocaine under a concurrent schedule of reinforcement with food available as the alternative. To test our hypothesis that cocaine would increase extinguished food-reinforcement response, the conditions were changed from a concurrent schedule to a simple FR schedule with only food available (and on the side of the operant panel that it has always been available). In all phases of the study, we recorded responses on both manipulanda, even though the discriminative stimulus (S^D) signaling cocaine availability was not illuminated. The results were further examined in terms of understanding the behavioral mechanisms of action leading to increased extinguished food-reinforced response.

2. Materials and methods

2.1. Subjects

Three socially housed adult male cynomolgus monkeys (Macaca fascicularis) with extensive (>4 years) cocaine self-administration histories served as subjects. Monkeys were housed in stainless-steel cages in a temperature- and humidity-controlled room maintained on a 14-hour light/10-hour dark cycle (lights on between 6:00 am and 8:00 pm) with ad libitum access to water. Monkeys were fed sufficient standard laboratory chow (Purina LabDiet) to maintain healthy body weights. Animal housing, handling, and experimental protocols were performed per the 2011 National Research Council Guidelines for the Care and Use of Mammals in Neuroscience and Behavioral Research and were approved by the Animal Care and Use Committee of Wake Forest University.

2.2. Apparatus

All monkeys were fitted with aluminum collars and trained to sit in a primate restraint chair (Primate Products). Behavioral experiments were conducted in ventilated, sound-attenuating primate chambers (1.5 × 0.76 × 0.76 m; Med Associates); descriptions of these chambers can be found in previously published work (Allen et al, 2024). Briefly, each chamber contained an operant panel with horizontally aligned photo-optic switches and stimulus lights (Model 117-1007; Stewart Ergonomics, Inc). In the middle of the operant panel, there was a food receptacle connected to a pellet dispenser (Med Associates), and there was a peristaltic infusion pump (Cole-Parmer Instrument Co) on the top of the chamber. Illumination of a red light inside the chamber accompanied food presentations.

2.3. Procedure: experiment 1: reinstatement of extinguished food-maintained response

All monkeys had extensive histories of responding under a concurrent schedule in which completion of an FR 30 on 1 switch resulted in the delivery of an intravenous injection of cocaine, and responses on the other switch produced 1.0 g banana-flavored food pellets. All monkeys had been responding on the concurrent schedule for >4 years. For this study, the response was only maintained by food pellet delivery under an FR 30 schedule. Under this schedule of reinforcement, only the white light above the response switch that signaled food availability was illuminated. Completion of the FR resulted in extinction (EXT) of the white light, illumination of the red light, and delivery of the food pellet; food pellet delivery was followed by a 10-second timeout. Sessions ended after 30 reinforcers were delivered or 60 minutes elapsed, whichever occurred first.

When responding under this schedule was deemed stable (mean reinforcer deliveries ± 20% for 3 consecutive sessions), the food pellet dispenser was unplugged. Monkeys could emit responses to the photo-optic switches, and the stimulus changes that previously occurred upon completion of the FR were still programmed, but food pellets were not delivered. Response was defined as extinguished when response rates declined by 80% of BL, with no trends in response across sessions. On average, this took 6 sessions. Once food-maintained response was extinguished, saline (0.5 mL, i.v.) or an acute dose of cocaine (0.03–0.3 mg/kg, i.v.) was administered 1 minute before the session. During these reinstatement sessions, the food pellet dispenser remained unplugged. Cocaine pretreats were typically tested 2 times a week—with at least 1 day between test sessions—and doses were tested in random order. Food-reinforced response was frequently re-established, followed by EXT and additional reinstatement tests.

Following the study of cocaine-induced increases in extinguished food-maintained responses, the effects of food pretreatments were examined. Monkeys received either 1 or 5 banana-flavored food pellets noncontingently while seated in the operant chamber 1 minute prior to the start of the session. Each test condition (1 or 5 food pellets) was determined at least twice, and the order of presentation was random. In monkey M-8181, only a 5-food pellet pretreatment was studied.

2.4. Experiment 2: active- and inactive-switch responses when discriminative stimuli signaling drug and food availability were manipulated

Prior to this experiment, food-maintained response was re-established and then extinguished. Following this, all monkeys received an acute pretreatment of 0.3 mg/kg (i.v.) cocaine 1 minute prior to the self-administration session. This dose was selected based on results from experiment 1. For this experiment, 2 manipulations were made. In one condition, no stimulus lights were illuminated during the session. This allowed us to determine the extent to which cocaine-induced increases in extinguished food-maintained responses were driven by the food S^D. Completion of the FR 30 on either switch had no consequence or stimulus changes. In the second condition, both stimulus lights were on over each switch. For this condition, we were interested in determining whether the S^D that had previously signaled cocaine availability but had not been illuminated for over a month would drive response to that switch. Under these conditions, completion of the FR 30 on either switch resulted in no consequences or stimulus changes. The monkeys were exposed to each of these conditions at least twice, and the order of presentation was random.

2.5. Data analysis

Individual subject data are shown as mean ± SD, and group data are shown as mean ± SEM. The main dependent variable for both studies was total responses to the active (ie, extinguished food-reinforced) and inactive (ie, previous cocaine-reinforced) switch. For experiment 1, paired t tests were used to confirm that food-maintained responses were extinguished after the food pellet dispenser was unplugged. Repeated measures ANOVA was used to compare the number of responses to the food-associated switch during EXT conditions and following cocaine pretreatments in each monkey to determine whether cocaine significantly reinstated extinguished food-maintained responses. Similarly, repeated measures ANOVA was used to determine whether the food pellet pretreatments significantly reinstated extinguished food-maintained responses. In addition to conducting analyses examining responses to the food-associated switch, these analyses were also conducted on the inactive (previous cocaine-associated) switch. For experiment 2, repeated measures ANOVA was used to compare responses between the active and inactive switches when both discriminative stimuli were off and when both discriminative stimuli were illuminated. For all statistical tests, significance was set at an α of .05, and all tests were analyzed with IBM Statistical Product and Service Solutions (SPSS). Furthermore, Bonferroni’s correction was used.

2.6. Drugs

(-)Cocaine HCl, supplied by the National Institute on Drug Abuse, was dissolved in a sterile 0.9% saline. All drug solutions used for intravenous administration were passed through a sterile 0.2 μm filter (Millipore Inc).

3. Results

3.1. Experiment 1: reinstatement of extinguished food-maintained response

When only the S^D signaling food availability was illuminated, the response was primarily to that switch (BL in Fig. 1A; Fig. 2A). Statistical analyses confirmed that when the food pellet dispenser was unplugged, food-maintained response was significantly reduced in all 3 monkeys (t = 20.24; df = 3 [t(3)], P = .002) (EXT in Fig. 1, A and B; Fig. 3) but did not influence response to the inactive switch (P = .781) (EXT in Fig. 2, A and B; Fig. 3). In monkeys M-8558 and M-8507 (Fig. 1A), 0.3 mg/kg cocaine significantly reinstated extinguished food-reinforced response (t[1] = −676, P < .001 and t[1] = −511.00, P < .001, respectively). In monkey M-8181 (Fig. 1A), 0.1 mg/kg, 0.17 mg/kg, and 0.3 mg/kg cocaine significantly reinstated extinguished food-reinforced response (P < .001, P < .001, and P = .042, respectively).

Fig. 1.

(A) The total number of active responses (ordinate) at BL following EXT and as a function of cocaine priming dose (red circles) and food priming (yellow triangles) (abscissa) in individual monkeys. Each point represents the mean ± SD from the last 3 sessions a condition was available. The asterisk (∗) indicates the dose/condition at which the number of responses was significantly higher (P < .05) compared with EXT conditions. Group data can be seen in (B) and represents the mean ± SEM. 1-P, 1 food pellet; 5-P, 5 food pellets.

Fig. 2.

(A) The total number of inactive responses (ordinate) at BL following EXT and as a function of cocaine priming dose (red circles) and food priming (yellow triangles) (abscissa) in individual monkeys. Each point represents the mean ± SD from the last 3 sessions a condition was available. The asterisk (∗) indicates the dose/condition at which the number of responses was significantly higher (P < .05) compared with EXT conditions. Group data can be seen in (B) and represents the mean ± SEM. 1-P, 1 food pellet; 5-P, 5 food pellets.

Fig. 3.

Average total responses during BL and EXT to the active “food” (orange bars) and inactive “cocaine” switches (black bars). Each bar represents the mean ± SEM from 3 monkeys. ∗P < .05.

Response to the inactive switch that was previously associated with cocaine availability was also examined (Fig. 2A). Although there was some response, no cocaine dose significantly increased response to the inactive switch in M-8558 (P > .05) or M-8181 (P > .05). In monkey M-8507 (Fig. 2A), 0.17 mg/kg cocaine, but no other cocaine dose, significantly reinstated response to the inactive switch (t[1] = −85, P = .007). Noncontingent presentation of 1 food pellet pretreatment did not significantly reinstate extinguished food-reinforced response in the 2 monkeys tested (P > .05), while noncontingent presentation of 5 food pellets significantly (P = .037) increased response in all 3 monkeys (Fig. 1A). Noncontingent food presentations had no effect (P > .05) on response to the inactive switch (Fig. 2A).

3.2. Experiment 2: active- and inactive-switch responses when discriminative stimuli signaling drug and food availability were manipulated

For experiment 2, individual subject data can be seen in Fig. 4 and group data in Fig. 5. Statistical analyses of group data revealed a significant main effect of the condition (F[3,6] = 8.64, P = .013) following 0.3 mg/kg cocaine pretreatments. Post hoc tests demonstrated that when both lights were illuminated, noncontingent cocaine induced a significantly (P = .007) higher number of responses to the previous cocaine-associated switch (mean [M] = 415.72, standard error of the mean [SEM] = 26.79) compared to the extinguished food-reinforced switch (M = 241, SEM = 35.52). However, when both lights were off, there were no significant differences between responses to the active versus inactive switch following 0.3 mg/kg cocaine injections (P = .41). Responses following 0.3 mg/kg cocaine injections were marginally higher (P = .064) for the active switch when the lights were on compared with when the lights were off (M = 95.33, SEM = 43.09). Also, the cocaine-induced response was marginally higher (P = .093) to the inactive switch when the lights were on versus off (M = 204.17, SEM = 95.43) (Fig. 5).

Fig. 4.

Total average responses during lights-on and lights-off conditions to the active “food” (orange bars) and inactive “cocaine” (black bars) switches in individual monkeys. Each bar represents the mean ± SD from the test sessions where a condition was available for at least 2 sessions.

Fig. 5.

Average total responses during lights-on and lights-off conditions to the active “food” (orange bars) and inactive “cocaine” (black bars) switches across all monkeys. Each bar represents the mean ± SEM. ∗P < .05.

4. Discussion

Earlier research indicated that stimulants, such as cocaine, could increase low-rate behavior at doses that would decrease high-rate response (Dews and Wenger, 1977). One of the exceptions to this rate-dependent effect involves low rates of response due to the removal of the reinforcing stimulus (ie, response EXT). Although a significant amount of previous data suggests that cocaine reinstates extinguished cocaine self-administration, the behavioral mechanisms underlying this phenomenon are not well understood (Spealman et al, 1999; Kalivas and McFarland, 2003; Odum and Shahan, 2004; Farrell et al, 2018). Thus, the goal of the present study was to further examine the impact that a history of cocaine self-administration would have on the ability of noncontingent cocaine to increase extinguished food-reinforced response.

For these studies, 3 monkeys with an extensive history of self-administering cocaine under a concurrent food versus cocaine choice schedule of reinforcement were used. At the start of this study, the contingencies were changed, and only food-reinforced responses were examined. In experiment 1, following response EXT, the highest dose of noncontingently administered cocaine (0.3 mg/kg, i.v.) increased extinguished food-reinforced responses in all 3 monkeys. In only 1 monkey, and at 1 dose, did cocaine increase response to the previous cocaine-associated side. Further examination of the role of discriminative stimuli found that when no lights above either switch were illuminated, response occurred at both switches but at low rates. This finding suggests that cocaine-induced increases in extinguished food-reinforced responses were influenced by the presence of a conditioned, albeit discriminative, stimulus associated with food presentation. However, when both discriminative stimuli were illuminated, the response was significantly higher on the previous cocaine-associated side compared with the food-associated side, supporting the powerful role of cocaine’s S^D effects in the reinstatement of cocaine seeking.

In an earlier study, Banks et al (2007) tested the hypothesis that cocaine-induced reinstatement in drug self-administration studies was analogous to substitution studies using drug discrimination assays. Banks et al (2007) first showed that cocaine reinstated cocaine self-administration in a group of monkeys with an extensive (>2 years) cocaine history. In another group of monkeys who were initially cocaine naïve, noncontingent cocaine did not increase extinguished food-reinforced responses, and 5 sessions of cocaine self-administration and cumulative cocaine intakes between 0.27 and 0.58 mg/kg (enough to show that cocaine had reinforcing effects) were not enough to change those effects. The present study tested the hypothesis that a long history (>4 years) of cocaine self-administration would alter the ability of noncontingent cocaine to increase extinguished food-reinforced response.

Consistent with our hypothesis, noncontingent cocaine administration increased extinguished food-reinforced responses in all 3 monkeys. To determine the behavioral mechanism(s) of action (Thompson, 1984), we took advantage of the previous experimental contingencies – namely, monkeys self-administered cocaine in the same chamber where food-maintained responses were studied. When only the S^D above the food-reinforced switch was illuminated, noncontingent cocaine only increased response on that operandum, not on the switch that had been associated with past cocaine reinforcement. However, when no discriminative stimuli were present, noncontingent cocaine resulted in increases in response to the previous cocaine-associated switch. Interestingly, when both sets of discriminative stimuli were illuminated, response to the cocaine-associated switch was higher than to the extinguished food-reinforced switch even though cocaine self-administration had not been studied in over a month, and during the test session, no reinforcers were presented for responding to either side. A limitation that should be acknowledged is that in experiment 2, the effect of an acute saline pretreatment was not tested. Thus, it is unclear if a saline pretreatment would also result in an increase in responses to the inactive (previously cocaine-associated) switch.

It is also important to note that many previous studies have shown that priming subjects with a drug increases extinguished drug response, commonly referred to as reinstatement (De Wit and Stewart, 1981; Spealman et al, 1999; Kalivas and McFarland, 2003; Farrell et al, 2018). Although the reinstatement paradigm has been widely implemented in the substance use field as a model of craving and relapse, there are notable concerns to be addressed when making the claim that reinstatement is highly translational in measuring relapse and drug craving. Although the reinstatement model is high in face validity, Katz and Higgins (2003) suggest that it largely lacks predictive and construct validity as a model for relapse and drug craving. Given that in this study, noncontingent cocaine administration reinstated previously extinguished responses maintained by delivery of food pellets, it is possible that reinstatement studies are not measuring relapse of “drug-seeking.” This premise requires further investigation and should be considered when designing experiments aiming to assess relapse.

The maintenance of drug self-administration involves not only the reinforcing effects of the drug but also the direct effects of the drug on ongoing behavior (Zernig et al, 2007). Taken together, the present findings suggest that a long history of cocaine self-administration can influence the direct behavioral effects of cocaine in addition to the reinforcing effects. This is of potential significance for several reasons. First, it is another indication that drugs do not have “properties” but rather produce effects through dynamic interactions between the organism and the environment. While the most obvious consequences of a chronic drug history are tolerance and/or sensitization to a particular effect mediated through pharmacokinetic and/or pharmacodynamic mechanisms, long-term drug exposure can also induce behavioral changes.

An important implication for the present study is that long-term cocaine self-administration resulted in increases in previously extinguished behavior that was not directly related to cocaine self-administration. Clinically, one could speculate that these previously nonreinforced behaviors would re-emerge with cocaine exposure, which could be an unanticipated challenge when designing effective and safe treatment plans for individuals with cocaine use disorder. Negative consequences could lead to social isolation and subsequent relapse to drug use. In the spirit of Dew’s (1955) study, the effects of self-administered cocaine on the brain and behavior cannot be determined in the absence of systematically studying ongoing behavior.

Conflict of interest

No author has an actual or perceived conflict of interest with the contents of this article.