Introduction

Early demonstrations that drugs could serve as reinforcers, maintaining operant behavior in laboratory animals, led to the development of a model of human substance use disorder ( Box 6.1 ). The traditional self-administration model was developed within a behavioral analysis conceptual framework that views drugs as reinforcers similar to other “natural” reinforcers such as food. The fundamental principle underlying behavioral analysis is that certain aspects of behavior are controlled by their consequences. A drug is said to be functioning as a reinforcer if responding for it is maintained above responding for saline or other control conditions. The traditional model entails training an animal to self-administer a drug during a short daily session, typically 1–2 h. A low ratio requirement is typically used, such as a fixed ratio 1, where each response produces a drug delivery. Intake is stable under these conditions, which allows for the determination of the effects of pharmacological and environmental manipulations on a stable baseline.

Although the rat is most often used in these studies, this model has been implemented with a variety of species including nonhuman primates, mice, dogs, cats, and baboons. A variety of operant responses have also been used, and typically they depend on the species being studied. For example, a lever press or a nose poke response is typically used for rats and mice, whereas a panel press response is typically used for nonhuman primates. The most common routes of administration are intravenous and oral, but intracerebroventricular, intracranial, inhalation, intragastric, and intramuscular routes have also been used. Generally, these studies use the route of administration that is most similar to the route used in humans for that particular drug. For example, animal studies with alcohol typically use an oral route of administration, whereas an intravenous route is typically used for drugs that have a rapid onset in humans, such as cocaine, methamphetamine, heroin, and nicotine. There is also growing interest in the development and use of inhalation self-administration procedures for the latter type of drugs, since, unlike the intravenous route, this route would not require the use of an indwelling catheter. Such approaches have been used successfully in nonhuman primates, with more recent work demonstrating its feasibility in rats and mice.

Historically, male animals have typically been used in drug self-administration studies. This focus was initially justified by higher rates of drug use and substance use disorder in men versus women. However, gender differences have narrowed over time, and among current adolescent populations, rates of drug use and substance use disorder are often similar between males and females. There are also important differences between men and women with respect to many aspects of substance use disorder, including initiation of use, the development of substance use disorder, and relapse and treatment. In addition, sex differences are observed in animal models of substance use disorder, indicating a biological basis for the gender differences observed in humans. Such differences also further support the need to include both sexes in studies on drugs as reinforcers and substance use disorder, a focus now mandated by the National Institutes of Health.

Results from animal drug self-administration studies have revealed good correspondence between humans and animals; drugs abused by humans generally maintain responding in animals, whereas drugs that do not maintain responding in animals are typically not abused by humans, indicating this paradigm’s utility for determining abuse liability. In addition, similar patterns of drug intake have been reported in humans and animals for ethanol, opioids, nicotine, and cocaine self-administration. These parallel results between the human and animal drug literature validate the animal model of drug self-administration and suggest that its use may lead to a better understanding of human drug-taking behavior and substance use disorder.

In addition to screening drugs for abuse liability, the traditional self-administration procedure has been used to study, through biochemical and pharmacological manipulation, the neurobiological processes underlying the drug reinforcement process. For example, by demonstrating that lesions in some areas of the brain decrease or abolish self-administration behavior, we have developed an understanding of the neuroanatomical substrates for drug reinforcement (e.g., Wise and Bozarth ).

Assessing Reinforcing Efficacy

Despite the advances in our understanding of drug reinforcement in animals, reinforcing efficacy, or a drug’s reinforcing strength, has been difficult to measure. The ability of a drug to support self-administration in laboratory animals under different experimental conditions is a measure of the drug’s strength as a reinforcer. Thus, a highly efficacious drug will be self-administered under a variety of experimental conditions such as low-dose conditions, conditions that require a large work effort, or enriched environmental conditions where other reinforcers are available as choices. In contrast, a weakly efficacious drug will be self-administered only under limited conditions such as food-restricted conditions, moderate- to high-dose conditions, conditions that require a low work effort, or impoverished environmental conditions where there are few or no other reinforcers available as choices. Such effects also depend on the route of drug self-administration. For example, under oral self-administration conditions, enriching the environment with toys or social peers can markedly reduce levels of opioid or psychostimulant drug intake, whereas, under intravenous self-administration conditions, such environmental manipulations are much less effective. Although it is generally believed that the reinforcing strength of a drug is related to its abuse liability, actually measuring reinforcing strength is not straightforward because factors other than the drug’s reinforcing effects can, directly and indirectly, influence responding (i.e., satiating effects, direct effects on responding, and aversive effects). The fixed-ratio schedule is typically used in studies investigating drug reinforcement in animals (e.g., 1–2 h sessions), and under these conditions, an inverted U-shaped relationship has been described between drug dose and rate of responding. That is, as dose increases, responding initially increases (ascending limb) and then decreases (descending limb). At low doses, responding decreases and these doses may not maintain responding. However, doses on the descending limb, which would be presumed to be more efficacious than doses on the ascending limb, maintain quantitatively similar levels or even lower levels of responding than those maintained by doses on the ascending limb. This issue is problematic for the interpretation of changes in reinforcing efficacy in that it is difficult to determine the direction of the change. A number of approaches have been taken to address this issue, including the use of rate-independent approaches such as the progressive-ratio schedule, the threshold procedure, second-order schedules, and choice procedures. Reinforcing efficacy is more readily determined using these approaches, and as such, they have been useful for determining changes as a result of pharmacological or environmental manipulation, or changes over time with the development of substance use disorder.

Progressive-Ratio Schedule

The progressive-ratio schedule is commonly used to evaluate the reinforcing strength of self-administered drugs, particularly psychostimulants. With this schedule, the ratio requirement to obtain a delivery progressively increases within a session, and the final ratio completed, or breakpoint, is believed to be a sensitive measure of motivation to obtain the drug (for a review, see Arnold and Roberts ). In contrast to the fixed-ratio schedule, the dose-effect curve under the progressive-ratio schedule is linear, whereby responding is directly related to reinforcer magnitude: an increase in the unit dose of the self-administered drug corresponds to an increase in breakpoint. This linear relationship allows for a more straightforward determination of the direction of change in reinforcing efficacy than is allowed by more traditional self-administration procedures. Other strengths are that responding for a particular dose of drug can be incredibly stable from day to day within subjects and that there are considerable individual differences in levels of responding between subjects. Sensitivity to pharmacological and environmental manipulations and to individual differences are thus strengths of the progressive-ratio schedule. Sex differences and hormonal influences on drug self-administration behavior are good examples of its sensitivity to individual differences in that under simple fixed-ratio schedules, sex differences and hormonal influences are generally not revealed, whereas, under progressive-ratio schedules, these factors robustly influence breakpoints (for a review, see Perry et al. ). Another advantage of this schedule is that it can be used reliably across different pharmacological classes of drugs including psychostimulants, nicotine, opiates, synthetic cathinones or “bath salts,” and alcohol. It has also been used successfully in several different species including rats, mice, and nonhuman primates with parallel effects observed in laboratory studies in humans with substance use disorder. However, as with the more traditional self-administration paradigms, the satiating and behavioral disruptive effects of drugs can also impact responding under a progressive-ratio schedule, particularly during earlier parts of the sessions, with high doses of the drug, and under low or slowly increasing progressive-ratio schedules.

Threshold Procedure

Recent studies have used the threshold procedure to disentangle reinforcing efficacy from satiating and behavioral disruptive effects. With this procedure, animals are given access to a descending series of drug doses under a fixed-ratio 1 schedule using either a between-session (i.e., dose progressively decreases with each successive daily session) or within-session approach (e.g., multiple doses are available each session with doses decreasing every 10 min ). The goal with either approach is to identify the lowest drug dose, or the threshold dose, that maintains self-administration. Each animal’s motivation to obtain the drug and its preferred level of drug consumption are then determined using a behavioral economic analysis of the response/intake data. For example, at suprathreshold doses, animals regulate their drug intake, presumably to a preferred level, such that as dose decreases, levels of responding increase. However, as dose further decreases and threshold is reached, the behavioral cost of maintaining the preferred level of intake exceeds the maximum acceptable price, and the animal stops responding.

The threshold procedure is similar to the progressive-ratio schedule in that the effort required to obtain drug (i.e., the price) progressively increases and allows for a sensitive measure of motivation for the drug. The two procedures appear to measure different aspects of motivation for drug in that one measures motivation to maintain a preferred level of drug intake (threshold), and the other measures motivation to obtain a particular dose of the drug (progressive-ratio). Like the progressive-ratio schedule, the threshold procedure appears to be sensitive to individual differences, pharmacological and environmental manipulations, and to changes over time within animals. ^a

References 32, 93, 112, 166, 195, 200, 227.

Second-Order Schedules

Second-order schedules have also been useful for minimizing issues of satiety and other rate-limiting effects of drugs on responding. Much of the early work using second-order schedules was conducted with nonhuman primates and focused on conditioned or secondary reinforcement (for a review, see Schindler et al. ). With this type of schedule, a nondrug stimulus, usually a light or a tone, takes on the characteristics of a reinforcer by its association with the drug delivery. Second-order schedules of drug delivery allow for the study of more complex behavioral sequences than do traditional self-administration procedures. Second-order schedules have also been used in studies with rats and mice, and these studies have been useful for the investigation of drug-seeking behavior (i.e., responding for drug that occurs prior to drug availability or when the drug is no longer available) and its neurobiological mechanisms (e.g., Di Ciano and Kumaresan et al. ).

Like the progressive-ratio schedule, second-order schedules minimize the descending limb of the dose-effect curve, allowing for determination of changes in reinforcing efficacy as a result of pharmacological or environmental manipulation. Another advantage is that high rates of behavior can be maintained by the conditioned reinforcer with relatively few actual primary reinforcers delivered. Nicotine is a good example of a drug that is robustly self-administered under second-order schedules, whereas, under simple fixed-ratio schedules, it has been more difficult to establish that it functions as a reinforcer. In fact, even under more traditional self-administration paradigms, nicotine maintains more robust levels of responding when the drug deliveries are paired with a stimulus cue, such as a light. However, one disadvantage of this approach is that it is often difficult to separate the reinforcing strength of the secondary reinforcer from that of the primary reinforcer.

Choice Procedures

Choice procedures are an increasingly popular tool for examining the reinforcing efficacy of drugs of abuse, particularly for work conducted in nonhuman primates (for review, see Banks, Banks and Negus, and Negus and Banks ). Early studies employing choice procedures showed that laboratory monkeys chose to self-administer a reinforcing drug over its vehicle. The procedures used in choice experiments typically involve one of three types of experimental schedules: discrete trial schedules, concurrent schedules, or concurrent chain schedules. Sessions typically begin with a sampling period during which the subject can respond to obtain each of the available reinforcers (i.e., a high versus a low drug dose or saline, or drug versus some other reinforcer, such as food). Animals then choose between the different reinforcer options in a series of trials by completing the response requirement on the lever associated with a particular reinforcer. Response allocation, rather than response frequency, provides a measure of the drug’s reinforcing strength. This feature allows for the determination of reinforcing strength relative to behavior allocated toward an alternative reinforcer. As such, choice procedures are believed to mirror more directly the real-world situation where drug users allocate resources to obtain drugs rather than other nondrug reinforcers such as food and extracurricular activities. Indeed, most self-administration studies conducted in humans with substance use disorder use choice procedures, where subjects choose between drug deliveries and a nondrug alternative such as money.

Studies have shown that laboratory animals not only choose drug over saline deliveries, but also prefer higher doses of drugs. For example, Carroll conducted a study in which monkeys chose between a standard dose of phencyclidine (0.25 mg/kg) or one of several other doses that were concurrently available (0.06, 0.12, 0.50, or 1.00 mg/kg), and found that subjects chose the large concentrations more often than the smaller ones. Similar results have been shown for a variety of other drugs including cocaine, ethanol, methadone, remifentanil, methylphenidate, and pentobarbital. ^b

b References 9, 104, 108, 116, 143, 153.

Modeling Phases of Substance Use Disorder

The majority of the preclinical studies on substance use disorder have used the traditional self-administration paradigm or other conditions that limit drug intake—that is, maintenance conditions that produce stable and relatively low levels of self-administration. As such, the behavioral and neurobiological principles defined by these studies may be restricted to drug reinforcement and not necessarily characteristic of substance use disorder. Specifically, although the positive reinforcing effects of drugs are believed to be a primary mechanism mediating drug use during the early phases of substance use disorder (i.e., during substance use initiation and maintenance), other characteristics appear to be critical for motivating drug use at later stages (i.e., following the development of substance use disorder, during relapse), such as a loss of control over use, compulsive use, and relief from craving. Several procedures have been developed to try to incorporate features of human substance use disorder that are not represented in more traditional procedures. These methods have focused on addressing critical questions regarding addiction, such as: “Why do some individuals develop substance use disorder but not others?” “What factors influence the transition from controlled or causal use to the development of substance use disorder?” and “What factors influence relapse or reinstatement of drug use?” The models that have been developed to address these questions are discussed below.

Animal Models of Drug Use Initiation

The reinforcing effects of a drug appear to be a primary determinant of vulnerability during the initiation or acquisition phase. Retrospective reports from individuals with substance use disorder reveal that the response to initial drug exposure varies from highly positive to negative, and some evidence suggests that individual differences in sensitivity to drug reinforcement are predictive of later use. Consistent with these clinical findings, there is considerable variability in laboratory animals in their propensity to self-administer drugs. Animal models of the acquisition phase have been developed to identify biological and behavioral factors underlying individual differences in vulnerability to the reinforcing effects of drugs of abuse that may apply to prevention efforts in humans (for a review, see Campbell and Carroll ). However, the acquisition phase is difficult to study because it is typically brief and is characterized by a sudden shift from low to high levels of intake. Thus, methods that slow the acquisition process and increase intersubject variability are necessary to observe this transitory period. For example, acquisition of drug self-administration is optimally investigated in drug- and experimentally-naive animals that are maintained under food-satiated conditions (e.g., food restriction serves as a stressor that can greatly accelerate the acquisition process and obscure individual differences) and tested under low dose conditions (e.g., high doses are associated with not only reinforcing effects but also direct effects and aversive effects that may interfere with responding). Under these conditions, individual differences are maximized, and some rats will acquire self-administration, whereas others will not; the question that is addressed is: “Which animals can detect the reinforcing effects of this low drug dose?”

A simple method of evaluating acquisition is to give an animal access to a drug during a daily experimental session, with deliveries available contingent upon an operant response (i.e., lever press; e.g., Davis et al. ). Another method that has been used to investigate individual differences in acquisition of drug self-administration is an autoshaping procedure. This procedure was adapted to the study of the acquisition of drug self-administration from methods used to study the acquisition of food-reinforced responding. With this method, daily sessions begin with a noncontingent drug administration component, wherein rats receive computer-automated infusions delivered on a random interval schedule that are paired with light cues and lever retraction. A self-administration component then follows wherein the lever remains extended and responses on it result in drug infusions. With both procedures, acquisition of drug self-administration is measured as the number of sessions needed to reach a criterion level of intake, which can be standardized and adjusted for dose and drug availability. The ratio of active to inactive lever-press responses is often used in conjunction with the intake criteria. All of the animals are included in the analyses, whether or not they acquire self-administration, and the focus is on how rapidly this process takes place and what percentage of each group of animals acquires drug-reinforced responding.

These acquisition methods have revealed a number of organismic and physiological factors that predict vulnerability to drug self-administration, such as genetic strain, impulsivity, exploratory behavior in a novel environment, reactivity to stress, innate saccharin preference, dopamine release and receptor levels in brain regions associated with drug reward, reactivity to injections of drugs age, and sex. For example, we used an autoshaping procedure to train male and female rats to lever press for either cocaine infusions (0.2 mg/kg) or heroin infusion (0.015 mg/kg) under a fixed-ratio 1 schedule (i.e., one response per infusion ). Under these low-dose conditions, female rats acquired cocaine and heroin self-administration at a faster rate than male rats, and a greater percentage of female rats acquired cocaine self-administration than did male rats. Similar results of enhanced vulnerability in females have also been reported in several other studies for cocaine and heroin, and for other drugs including nicotine and phencyclidine. As mentioned previously, however, individual differences can be obscured under conditions that speed up the acquisition process. For example, a recent study showed that when rats are food restricted, which serves as a stressor and enhances rates of acquisition, sex differences in rates of acquisition are reversed, and male rats acquire both cocaine and nicotine self-administration faster than female rats.

Environmental factors, such as feeding condition, the presence of an alternative nondrug reinforcer, exercise, and drug history, can also greatly impact acquisition. For example, several studies have reported that exercise via running in a wheel attenuates the acquisition of drug self-administration for numerous drugs of abuse including nicotine, cocaine, heroin, and methamphetamine (for review see Lynch et al. and Lynch et al. ). The effects of exercise on acquisition are robust, and have been observed for both concurrent conditions, that is, when exercise is available during the self-administration session, and for nonconcurrent conditions, that is, when exercise precedes or follows the daily self-administration session, indicating that its effects extend beyond its ability to function as an alternative nondrug reinforcer. Rates of acquisition also vary widely as a function of drug dose, type of drug, and route of administration. For example, under high-dose conditions with a drug that rapidly enters the brain after an intravenous infusion (i.e., cocaine), most, if not all animals, will acquire self-administration rapidly. However, when lower intravenous doses of the drug are used, or an oral route of administration is used, fewer animals will acquire, and the rates of acquisition become much slower. Similarly, with drugs such as caffeine or alcohol, which are considered to have a less-intense or less-rapid onset of action, the acquisition process is slowed. With oral administration, the taste of the drug can also influence the probability and rates of acquisition (e.g., the acquisition of oral alcohol self-administration is relatively slow because animals typically have an aversion to the taste of unsweetened alcohol).

Although these studies provide insight on factors that influence the initial vulnerability to drug use and reinforcement, they do not necessarily reflect vulnerability factors for the development of substance use disorder. For example, in humans, although most individuals have used drugs (i.e., greater than 90% if alcohol is included ), only a small percentage will develop a substance use disorder. Thus, recreational drug use does not invariably lead to the development of substance use disorder. This is also true for preclinical studies, with results suggesting that drug access conditions are critical for determining whether an animal will develop features characteristic of substance use disorder.

Animal Models of Substance Use Disorder

Two of the defining features of substance use disorder in humans—loss of control over drug use and the resulting excessive use of the drug—have been modeled in animals using several different methods (for a review, see Edwards and Koob, Lynch, and Roberts et al. ). For example, early studies showed that when nonhuman primates or rats were given unlimited 24-h/day access (i.e., each response is reinforced under a fixed-ratio 1 schedule) to intravenous infusions of psychostimulants such as cocaine, d -amphetamine, or methamphetamine, they self-administrated high levels of the drug and displayed binge-abstinent patterns of intake with drug self-administration occurring erratically throughout each 24-h period. Unlimited access to opiate drugs, such as heroin or morphine, also results in high levels of intake that increase, or escalate, rapidly over time. Toxicity also develops rapidly under these conditions, particularly for psychostimulants and opiates, thus necessitating the use of procedures that limit access to these drugs in some way.

More recent studies have attempted to capture these features—excessive intake and a dysregulated or escalating pattern of drug use—but without serious signs of toxicity. For example, excessive drug intake with limited signs of toxicity has been observed under unlimited 24-h/day access conditions with low unit doses of drug and under unlimited access conditions that restrict the number of hours of access each day (i.e., 6–12 h daily ) or each period of continuous access (i.e., 72 h ). The long access procedure developed by Ahmed and Koob is probably the most well-established extended-access procedure. With this procedure, animals, typically rats, are given unlimited access (fixed-ratio 1) to intravenous infusions of a drug for 6–12 h/day. Toxicity is limited under these conditions, and animals self-administer high levels of the drug and show an escalating pattern of drug intake over time. Escalation of intake has been observed under these conditions for numerous drugs of abuse including cocaine, methamphetamine, methylphenidate, heroin, fentanyl, oxycodone, and bath salts. In addition, drug-use escalation has been observed in nonhuman primates and mice, and for other routes of drug self-administration, including oral and more recently, vapor self-administration of sufentanil.

Another method that allows for extended access to the drug with limited toxicity is the discrete trial procedure, wherein animals are given 24-h access to drug infusions that are available in discrete 10-min trials. With this method, excessive intake and binge-abstinent patterns of use are observed as access conditions increase. For example, under short-access conditions (1–2 discrete trials/h; 1 infusion/trial), rats consume low and stable levels of drug and show a diurnally regulated pattern of intake (i.e., intake occurs predominately during the dark phase). However, under extended-access conditions (i.e., four discrete trials/h, 1.5 mg/kg/infusion), rats self-administer high levels of drug in “binge-abstinent” patterns, taking nearly every infusion available for the first 1–2 days, followed by periods of self-imposed abstinence interspersed with periods of active drug use. Although most of the work conducted with this procedure has focused on cocaine self-administration, similar findings of high intake and binge-abstinent patterns of use have also been reported for heroin and the combination of cocaine and heroin self-administration. A variation of this procedure was developed recently that allows multiple infusions within each trial, thus allowing animals to choose their preferred “dose.” With this model, the intermittent-access model, rats have unlimited fixed-ratio 1 access to cocaine or methylphenidate infusions within 5-min discrete trials that initiate every 30 or 60 min for up to 24-h/day. Under these conditions, rats self-administer high levels of the drug in repeated binge cycles and show an escalation in total daily levels of drug intake as well as dose escalation within trials. One advantage of this procedure over the other methods is that it appears to induce both escalation of intake and binge-abstinent patterns of use.

It is important to note that other critical features of substance use disorder, including increased motivation for the drug, use despite negative consequences, and increased drug-seeking and relapse vulnerability, are also observed following extended-access self-administration, particularly when behavior is examined after an abstinence period. For example, a history of escalating cocaine, methamphetamine, and heroin self-administration under the long-access procedure has been shown to lead to enhanced drug-seeking following protracted abstinence (i.e., 7 days or more) as compared to earlier time-points during abstinence or to short-access controls. It also induces incubation of drug-seeking, or the progressive increase in drug-seeking over abstinence, as well as persistent responding despite concomitant punishment (e.g., drug deliveries are paired with shock ), a feature believed to reflect use despite negative consequences. Extended-access drug self-administration under the discrete trial procedure also induces high levels of drug-seeking and its incubation over abstinence, as well as an enhanced motivation for the drug. ^c

c References 62, 125, 134, 140, 156, 183, 184.

Other drugs, such as nicotine and ethanol, typically can be available under unlimited-access conditions with limited toxicity, and results from studies with these types of drugs have also revealed “addiction-like” behavioral phenotypes. For example, Wolffgramm and Heyne developed an animal model of this transitional phase for oral alcohol, etonitazene, and amphetamine self-administration in rats. Their procedure entails long-term ad libitum self-administration (1–2 months) followed by an extended drug abstinence period (4–9 months). Subsequently, rats were retested on self-administration behavior, and those animals that developed escalating patterns of intake prior to abstinence, self-administered higher levels of intake compared with rats that did not show escalation. These animals were also resistant to punishment and continued to self-administer the drug even after it was adulterated with bitter-tasting quinine. Similar effects can also be observed under shorter-term self-administration conditions and following shorter periods of abstinence when access conditions are cyclical and alternate between self-administration and abstinence.

Access conditions, drug dose, route of administration, and the drug being self-administered are crucial factors for the observation of excessive and dysregulated patterns of self-administration and the subsequent development of an addicted phenotype. The time-course for the development of certain features of substance use disorder also appears to differ between the different extended access procedures. For example, several recent studies have shown that extended access under the intermittent-access procedure reliably induces an enhanced motivation for the drug as soon as 18 h after the last self-administration session. Seven days appears to be a threshold condition for the induction of this motivational shift following self-administration under the discrete trial procedure. The time-course also appears to be prolonged following self-administration under the long-access procedure, with results from several studies showing that motivation is unchanged from baseline even when assessed after 5 or more days of abstinence. This interpretation is also consistent with results showing that key molecular correlates for the incubation of drug-seeking emerge 3–4 weeks after self-administration under the long-access procedure.

Individual differences during this transition phase also have been reported. For example, female rats require less drug exposure and/or shorter periods of abstinence than male rats to display an increased motivation for cocaine, and in female rats, estradiol may be required for its development. A high preference for sweets, level of anxiety, and level of reactivity to stress, drug injections, and novelty, as well as a preference for drug over food, have also been reported to predict a vulnerability for the development of an addicted phenotype including drug escalation/dysregulation, enhanced motivation, and drug-seeking behavior. ^d

d References 13, 64, 98, 122, 145, 170, 172.

e References 12, 19, 26, 62, 69, 79, 152, 183, 207.

Animal Models of Relapse

Relapse, or recurrent resumption of drug use after abstinence, is one of the most challenging problems in the treatment of substance use disorder. Various types of stimuli can precipitate relapse, including internal cues such as reexposure to small “priming” doses of the drug or stress and external cues such as specific people and places that were associated with drug use. Often, external stimuli lead to drug use, and then internal stimuli sustain relapse. Animal models of relapse have been developed and have provided critical information on the neurobiological mechanisms underlying vulnerability to relapse.

One model that has been used to investigate mechanisms underlying drug-seeking and relapse vulnerability over abstinence is the reinstatement paradigm. With this procedure, animals are trained to self-administer a drug and, once stable, responding is extinguished by discontinuing drug delivery. After responding reaches some criterion of unresponsiveness (e.g., fewer than 10 responses/session or h), the ability of various stimuli to reinstate drug seeking is determined under conditions of nonreinforcement (i.e., responding is no longer reinforced by the drug). Extinction and reinstatement testing typically occur following the self-administration component during abstinence either within a single session (e.g., six 1-h extinction sessions followed by a 1-h reinstatement session all run with 1-day ]), or over multiple daily sessions (e.g., daily 1-h extinction sessions followed by a 1-h reinstatement session conducted on a separate day ). This sequence of events, self-administration, extinction, and reinstatement, can also occur all within a single session (e.g., de Wit and Stewart and de Wit and Stewart ). Results from preclinical studies using the reinstatement procedure have revealed that, like in humans, in laboratory animals, multiple stimuli can trigger drug-seeking behavior including priming doses of the drug, stress, and drug-associated cues (for a review, see Katz and Higgins ). These parallel findings validate the reinstatement model, and indicate its utility for screening potential interventions for relapse prevention in humans as well as for studying factors influencing and mechanisms underlying relapse.

Although most of the early studies using the reinstatement paradigm determined effects following short-access self-administration, more recent work has focused on determining effects following extended-access self-administration, given that levels of drug-seeking are higher and involve different neurobiological mechanisms following extended versus short-access self-administration. ^f

f References 62, 70, 113, 115, 126, 146, 179, 190, 194.

g References 130, 136, 164, 174, 175, 191, 192.

A number of factors are known to predict a vulnerability to seeking drugs following abstinence from short-access self-administration including high preference for saccharin, high responsiveness to acute and chronic drug exposure and novelty, risky decision making, a pattern of drug intake prior to reinstatement testing, age, and sex. Although less work has focused on identifying factors predictive of vulnerability following extended-access self-administration, results from the available studies indicate that similar factors are involved including saccharin preference, impulsivity, avoidance of a drug-paired taste cue, age, and sex. Notably, there appear to be important interactions of cues used to trigger reinstatement responding and vulnerability factors. For example, although female rats show enhanced reinstatement responding compared with male rats following exposure to priming injections of a drug, male and female rats have been reported to respond at similar levels following exposure to drug-associated cues. Similar results have been reported in laboratory studies in men and women with substance use disorder (for a review, see Lynch et al. ), suggesting that vulnerability to relapse may be due to a complex interplay of environmental and biological factors.