From Pellets to Prizes: Preclinical and Clinical Perspectives on Contingency Management for the Treatment of Substance Use Disorder

Abstract

Despite substantial evidence of its efficacy, contingency management as a treatment for substance use disorder remains widely underutilized. Lack of long-term efficacy, inability to address the cause of addiction, and lack of identifiable mechanism feature prominently among the concerns reported by both clinicians and the public. In this narrative review, we consider the possible mechanisms of change involved in treatment and recovery for substance use disorders, including intrinsic motivation, decision making, neural plasticity, and abstinence, with a particular focus on how contingency management may act as a vehicle for facilitating long-term change in people with a substance use disorder, employing our combined expertise as a preclinical behavioral neuroscientist and a licensed clinical social worker specializing in the translational behavioral science and treatment of substance use disorders respectively. We present a theoretical model that combines the behaviorist foundations of contingency management with recent preclinical and clinical data to explain how the use of external reinforcers might facilitate long lasting, intrinsic change in behavior that addresses the pathology of addiction. Ultimately, we suggest that the use of extrinsic reinforcers during the application of contingency management might reinforce immediate behaviors that permit the development of longer-term mechanisms of change, such as abstinence, inhibitory control, or self-efficacy, and that when extrinsic reinforcement is removed at the conclusion of a contingency management program, recovery-related behaviors might still be maintained by these mechanisms.

Introduction

In 2021, 46.3 million people in the United States alone were diagnosed with a substance use disorder (SUD) and only 6.3% of them sought treatment.1,2 SUD is a chronic, relapsing condition3 characterized by loss of control and significant social, occupational, and physical impairment. SUD is considered “chronic” due to the persistent alterations in brain structure and function and “relapsing” due to the high rate of return to substance use during or after treatment. In general, longer-term care, in keeping with care provided for other chronic conditions, is associated with the best outcomes.4–6 Though relapse occurs in an estimated 40–60% of cases,7 remission rates remain promising; one study estimated lifetime probability estimates of remission for nicotine, alcohol, cannabis, and cocaine between 83–99% depending on the substance and reported that half of the participants remitted within 5–26 years of diagnosis.8 One important caveat in these reported remission rates is the likely exclusion of people for whom substance use had already resulted in death, incarceration, or institutionalization. Still, recovery is possible and treatment for SUD is available.

Models of Addiction

A universal definition of “addiction” remains elusive and, perhaps, impossible.9–11 Importantly, the clinical diagnosis of substance use disorders is made in reference to the presence and severity of behavioral symptoms, rather than cause.12 Unified theories of addiction have tended to adhere to one of several theoretical frameworks or “models” of addiction.13 For example, the self-medication hypothesis, which postulates that people use drugs as a coping mechanism for underlying psychological discomfort,14 is central to the psychological model of addiction. The oldest view, called the moral model, defines addiction as an individual choice while newer models that emphasize the neurobiological, genetic, social, and psychological contributors to addiction pathology and vulnerability, including the brain disease model, tend to reject the assertion that addiction is a “choice”.13,15–17 The most inclusive model, the biopsychosocial framework, employs a combined approach.15,17

For those that adopt them, these models serve not only as a framework for the cause of addiction, but also as a template for treatment. The language used to discuss addiction and people with SUD can influence health care workers’ attitudes toward patients.18 Although the brain disease model is prominent in clinical settings, the themes of the moral model are far from extinct.19 Conjecture regarding the etiology of addiction denotes conjecture regarding the appropriate treatment. Of course, that a treatment proves effective in producing clinically relevant outcomes, like abstinence, does not necessarily indicate that the treatment addresses the cause of addiction, in a similar fashion as taking a pain-relief medication masks pain without necessarily addressing the source of the pain.

The purpose of this review is to consider one form of SUD treatment, contingency management (CM), which employs external reinforcers to maintain treatment-relevant behaviors and appears effective in helping people in recovery maintain abstinence. The mechanism of this efficacy is unclear, however. Here, we offer a theoretical mechanism for CM which alludes to multiple models or etiologies of addiction, including choice, learned behavior, and neurobiological reactions to psychoactive substance use, in the hope that hypothesizing how CM might work will direct future clinical research and ultimately increase how often CM is used to aid people in recovery.

Treatment of SUD

Frontline formal treatments include Cognitive Behavioral Therapy (CBT) and Motivational Interviewing (MI).20 While often used together, these distinct treatment styles differentially address symptomology. CBT is an approach combining components of both cognitive and behavioral therapies to address how thoughts, emotions, and behaviors interact cyclically.21 The use of CBT in treatment for SUD often involves helping individuals build coping skills for addressing maladaptive thoughts and uncomfortable emotions that lead to self-defeating behaviors including the use of substances. Alternatively, MI is a style of treatment rooted in the transtheoretical model of change, which focuses on interventions to enhance motivation based upon an individual’s current level of motivation.22 Both CBT and MI are evidence-based therapies with demonstrated efficacy in reducing the symptoms associated with SUD compared to non-treated control groups.23,24

Many people with SUD enter recovery without the help of formal treatment,25,26 and some even return to recreational substance use.27 For those who choose not to or are unable to seek formal treatment, a common alternative is support or self-help networks including 12-step programs like Alcoholics and Narcotics Anonymous (AA/NA) which are also often integrated with formal treatment programs in the form of Twelve Step Facilitation (TSF). Historically, research on the efficacy of 12-step programs has been inconclusive and controversial. One early meta-analysis concluded that the outcomes and experiences associated with 12-step were too heterogenous and lacked statistical power for rigorous analysis.28 Another concluded that 12-step attendance produced worse outcomes than no treatment29 and noted that experimental results were likely compromised by selection bias. The issue with selection is also noted in subsequent reviews and meta-analyses.30,31 However, the most recent meta-analysis, conducted in 2020, concluded that AA/TSF attendance is at least as effective as other treatments, including CBT, for the treatment of AUD.31 Though not broadly considered an application of CM, 12-step programs do involve earning tokens representing abstinence milestones, in the form of small coins or chips. Whether these chips function as reinforcers is unclear, but their delivery is contingent upon self-reported abstinence at progressive intervals. The role of social reinforcement is also a prominent and often distinguished factor in TSF. Consistent with formal treatment models, AA/NA promote the idea that long-term remission is associated with long-term (often, lifetime) abstinence and meeting engagement.32

Contingency Management

CM refers to the use of the principles of operant (instrumental) conditioning33,34 to induce behavioral change. In the treatment of SUD, CM involves the use of external incentives (money, vouchers, prizes, etc.) to reinforce recovery-related behaviors including treatment adherence (eg attendance or engagement) or abstinence (eg negative urine, blood, or breath tests) which can be earned according to various schedules of reinforcement.35,36 For a more comprehensive review of the history and implementation of CM, we suggest the following reviews: Higgins et al 2004 and Petry et al 2017.37,38

Numerous individual studies and several comprehensive meta-analyses have consistently indicated that abstinence-contingent CM has a significant positive impact on abstinence.39–49 Some, but not all, studies also report a significant positive effect on treatment retention46,48 and drug craving.50 Non-abstinence contingencies are also effective at altering behavior. Attendance-contingent CM has a moderate effect on attendance and a small effect on abstinence51 quantified as a 54% decreased likelihood of relapse during the study period. CM is also efficacious in promoting treatment compliance with medications for SUD.52,53 CM can be administered as an adjunctive treatment alongside cognitive behavioral therapy (CBT) and some studies suggest that the use of CBT and CM together is more effective than either treatment alone42,54 while others find that CBT and CM, though both effective, do not have an additive effect on treatment outcomes.55 CM does, however, improve outcomes in participants already enrolled in TSF communities of care.56–62 There is some evidence that monetary reinforcers may be more effective compared to merchandise or prizes49,63 and that efficacy of CM is sensitive to both magnitude of the drug (dose) and magnitude of alternative reinforcer in lab-based studies.63,64 Yet, studies employing non-monetary reward have also reported significant positive results. Taken together, these studies suggest that CM is effective in various applications, including different contingent behaviors, as a primary or adjunctive treatment, and employing different modes of reinforcement.

Proponents of CM argue that its success in promoting abstinence is a valid mechanism of change.38 Indeed, the idea that abstinence is the primary mechanism of SUD recovery and treatments that prove effective in long-term remission do so by virtue of abstinence is a common notion across formal and informal SUD treatment communities.65–67 In favor of this premise is evidence that contingent-incentive (reinforcing positive behaviors apart from abstinence) models are less effective than abstinent-incentive models within the CM framework.51,68–71 Despite the demonstration of efficacy similar to, if not greater than, other formal methods of SUD treatment, CM remains underutilized.72 Among the concerns touted by the public and healthcare professionals are financial and logistic challenges, assertions that CM constitutes “bribery”, undermines agency or intrinsic motivation, and failure to address the pathology of addiction.72–75 One study of the beliefs of practitioners further revealed that “orientation to the 12-step philosophy” was negatively associated with acceptance of CM.73

People with SUD face significant stigma from the public and healthcare professionals alike, even compared to others with mental health conditions.76–79 Indeed, if abstinence is touted as both a goal and the mechanism of recovery, and CM is effective in promoting abstinence, then the criticism of CM cannot be purely based on a lack of efficacy. Addressing the logistical concerns surrounding the cost and practicality of CM implementation is beyond the scope of this review, others have recently addressed these issues (see: Lu et al 202480 and Rawson et al 202381,). Neither will we argue for the efficacy of CM, which has been demonstrated. This review will instead focus on concerns regarding the mechanism of CM, both that CM might fail to address the root cause of addiction and that CM undermines intrinsic motivation. We will address both the current clinical studies that provide direct evidence on the nature of behavioral change associated with CM and the preclinical and basic literature which may elucidate the hypothetical mechanisms by which these changes might be achieved; in other words, we consider the question of the mechanisms of CM from studies using sugar pellets to reinforce behavior in an experimental setting to clinical facilities using prizes to reinforce treatment-related behaviors in people with SUD.

Learning, Conditioning, and the Management of Contingencies

I shall devote the body of this paper to a description of experiments with rats. But I shall also attempt in a few words at the close to indicate the significance of these findings on rats for the clinical behavior of men. – Edward C Tolman82

Operant or instrumental learning refers to the relationship between behavioral actions and outcomes, such that manipulation of outcome (eg reinforcers) can induce alterations in behavior.33 Though traditionally credited to Burrhus Frederick Skinner, an earlier iteration of instrumental learning was studied by Edward Thorndike in his seminal puzzle box experiment.33,34 Thorndike’s experiment involved placing a cat in a box with a latched door that could be opened from the inside by pressing on a pedal attached to a wire loop, allowing the cat to escape the box and gain access to food placed nearby. Over multiple trials, the cats gradually began to escape the box more quickly. Thorndike originally theorized that what the cats were learning was an association between a situation and a response that was progressively strengthened by the availability of food upon escape. Several years later, Thorndike introduced the extension of this observation, called the “law of effect”, which postulates that behaviors followed by satisfying outcomes are strengthened and behaviors followed by unsatisfying outcomes are weakened.83–85 The law of effect may be considered the spiritual precursor to the theory of reinforcement and Skinner’s lifelong exploration of the principles of operant conditioning.

In the early 1930s, B.F. Skinner began the series of experiments that would become the foundation of his life’s work.33 Using his eponymous behavior chamber, Skinner observed that rats would press a lever when the delivery of a food pellet was made contingent upon the lever press. Skinner concluded that the food pellet “reinforced” lever pressing behavior, such that the delivery of the food pellet made the lever pressing behavior more likely to be repeated. Skinner believed behavior was learned via the presence of reinforcement (a stimulus, event, or consequence that increases the likelihood the preceding behavior will occur) or punishment (decreases) and was thus lawfully shaped by the external environment. Indeed, Skinner argued that there was little empirical use for theories, which he defines as “any explanation of an observed fact which appeals to events taking place somewhere else, at some other level of observation”, that attempt to explain behavior using reference to internal processes.86 In other words, Skinner felt there was more utility in describing observable behavioral patterns than postulating the unobservable or private behaviors (thoughts, feelings, etc.) underlying those patterns. Skinner’s strict rejection of “attempts at explaining behavior using internal processes”87 resulted in the designation of his views as “radical behaviorism”. It is important to note, however, that Skinner’s skepticism of the empirical measurement of internal states should not be misconstrued as a denial of internal processes. Rather, Skinner rejected the use of introspection as a dependent variable.

Skinner later described how the principles of reinforcement might extend to learning in a school setting:

Eventually, of course, the pupil may be able to check his own answers and achieve some sort of automatic reinforcement, but in the early stages the reinforcement of being right is usually accorded by the teacher.88

Here, Skinner describes a hypothetical but critical interchange between behavior that is initially externally reinforced (by the teacher) but becomes internally or “automatically” reinforced (by the learner). So, even a “radical” behaviorist interpretation of extrinsically reinforced behavior does not deny the existence of concomitant or subsequent intrinsic motivation.

Edward Tolman, a contemporary of Skinner and Thorndike, offered an alternative to radical behaviorism, “purposive behaviorism”, a bridge between behaviorism and cognition.87 Tolman’s well-known early work involved the use of selective breeding to demonstrate that maze learning in rats was partially informed by genetics.89,90 He demonstrated that rats bred from earlier generations of “maze-bright” or quicker to learn rats were themselves more adept at maze navigation, thereby implying that learning is not simply a function of the environment, but also dependent on individual factors. In 1930, Tolman conducted a more direct assessment of the role of reinforcement in learning.91,92 In this experiment, rats were separated in three groups and each rat was placed into a fourteen-arm T-maze once a day for two weeks. At the “end” of the maze was a platform upon which food might be placed. Tolman recorded the time rats in each group spent navigating the maze. The rats in the first group always found food available at the end of the maze; we will call them the always-reinforced group. The rats in the second group never found food at the end of the maze; we will call them the never-reinforced group. The rats in the third group found no food at the end of the maze for the first ten days but did find food for the remaining four days; we will call them the late-reinforced group. As you might expect, rats in the always-reinforced group demonstrated a steep learning curve: they quickly learned to navigate the maze and their time spent in navigation decreased each day with fewer errors (wrong turns). Rats in the never-reinforced group did not race to the end of the maze, such that neither their time spent navigating nor their error rate decreased significantly across the two weeks spent in the maze. Rats in the late-reinforced group displayed similar behavior to the rats in the never-reinforced group for the first ten days. However, after the introduction of food on eleventh day, their performance improved to the level of the rats in the always-reinforced group by the twelfth day. This latent learning effect, implied that the rats in the late-reinforcement group had been learning about the maze during the first ten days of the procedure but had no incentive to demonstrate this learning (via performance) until the eleventh day. In other words, the reinforcer gave them a reason to display the learned behavior (where before, there was little purpose in navigating to the platform with haste or precision). This was, Tolman argued, evidence that learning can occur in the absence of reinforcement, but that reinforcement provides motivation for the demonstration of that learning: reinforcement is the motivator, not the mechanism.

The same year that Tolman published these pivotal findings, Donald MacFarlane published the results of another maze learning study in rats.93 MacFarlane placed the rats in a maze filled with water, through which the rats could swim to a platform where food was available. Once the rats could successfully swim to the platform, MacFarlane drained the water and returned the rats to the same maze. The study was designed to assess whether rats, in learning to swim the maze, had learned a specific motor pattern or a broader behavioral strategy. The observation that rats could readily and successfully run through the maze to retrieve the food supported the latter supposition. In short, the rats demonstrated that what they had learned during the maze swimming was not the specific movements of swimming, but a flexible, goal-oriented spatial awareness.

The contributions of Tolman and MacFarlane are significant in our consideration of CM. People in recovery from SUD are learning. Both formal and informal treatment like CBT and TSF provide tools and strategies for maintaining abstinence; even those who do not seek formal treatment employ some strategy to adjust their behavior. CM provides scaffolding in the form of external reinforcement. Tolman demonstrated not that extrinsic reinforcement was required for learning, but that it generates motivation to use learning. Macfarlane demonstrated that what is learned during operant conditioning is broader than a simple motor pattern. People who engage in SUD treatment or attempt abstinence are learning flexible behavioral strategies for maintaining abstinence. Thus, if learning that occurs in the absence of reinforcement can be externally incentivized and learning involves the development of behavioral strategies, then reinforcement during CM may provide a reason to enact strategies, perhaps coping skills like those developing in CBT or TSF, developed in another setting. We argue that CM neither creates nor diminishes intrinsic motivation but creates the opportunity for its behavioral expression. Furthermore, CM creates an external reinforcement structure that might mitigate the pathological behavioral, cognitive, and neural changes associated with chronic substance use and SUD that present a significant challenge to people attempting to maintain abstinence.

Mechanisms of Recovery

Abstinence

There is no universally accepted definition for recovery from SUD, though major organizations like the National Institute of Drug Abuse (NIDA), the Substance Abuse and Mental Health Service Administration (SAMSHA), and the Betty Ford Institute include abstinence or sobriety as a central component.65–67 Some experts suggest that abstinence per se may be less important than improvements in psychological, social, and health-related outcomes which may not require complete abstinence to improve quality of life.94 Indeed, harm reduction interventions are effective in reducing the negative consequences associated with substance use.95 However, in the United States, most treatment is abstinence-based,96,97 and some studies suggest that abstinence is independently associated with better quality of life following recovery.98,99 Thus, while abstinence is not the mode of recovery, it is a prominent recovery goal and produces significant improvement in a variety of domains.99–101

Importantly, abstinence may be not only a benchmark for recovery, but also a mechanism. The achievement of long-term abstinence demonstrates a significant shift in behavior and the acute stages of abstinence are particularly difficult.102–106 However, that abstinence becomes easier over time provides some evidence that cessation from substance use may be corrective. As we will discuss below, the alterations in behavior, decision making, and neural structure and function that are associated with chronic substance use can be reversed or recovered following abstinence.

Abstinence from a substance can alter behavior, and even preference, for that substance. This effect can be observed beyond psychoactive substances, but in the realm of natural rewards. For example, maintenance of a low-fat, low-sugar, or low-sodium diets results in decreased preference for high-sugar, high-fat, and high-sodium foods.107,108 In the brain, the hedonic properties of reinforcers are encoded in several “hedonic hotspots”. Alteration of the activity of these regions, like the nucleus accumbens shell, can induce hedonic shift or change in the subjective appraisal or liking of a reward.109 Abstinence from drugs is associated with neural recovery that begins soon after cessation from substance use110,111 and is associated with recovery of neurocognitive function.112,113 Thus, restriction from a drug or non-drug substance may allow for recovery of neural systems dedicated to preference or pursuit of that substance.

If abstinence is the driving force behind recovery, then any intervention that effectively promotes abstinence should facilitate recovery. Of course, this is not always the case. For example, people who are incarcerated for a drug-related offense experience enforced abstinence and often have restricted access to formal treatment.114 Incarcerated people who receive treatment during or shortly after incarceration have better treatment outcomes than those who do not.114–116 Similarly, animals denied access to a previously self-administered drug will return to drug seeking and intake when the drug is again made available, even after long stretches of time (see below). Thus, abstinence may be a necessary (for some) but not sufficient step in long-term recovery from SUD.

Each of the mechanisms we will discuss below (extinction learning, intrinsic motivation, decision-making, and neuroplasticity) are associated with abstinence. We will begin by discussing how models of extinction learning and post-extinction return to substance use demonstrate that abstinence alone may not fully explain why CM is efficacious in humans. Next, we will address the role of intrinsic motivation in maintaining abstinence. Then, we will consider how the decision-making inherent in the maintenance of voluntary abstinence is challenged by drug-associated differences or alterations in impulsivity, delay discounting, and response inhibition. Finally, we will review the neural effects of chronic substance use, the challenge to abstinence posed by those effects, and the evidence for recovery following cessation. At each stage, we will discuss how CM may provide a structure that promotes voluntary abstinence using extrinsic reinforcement, allows for the development of intrinsic motivation via self-efficacy, enables the strengthening of effective strategies for promoting abstinence, supports changes in decision making, and thereby enables neural recovery.

Extinction, Renewal, and Resurgence

The drugs involved in substance use disorders are readily self-administered in lab settings and function as reinforcers for operant learning.117–119 Cues associated with these drugs can become classically conditioned, such that exposure to such cues will induce expectancy for the drug.119 Thus, recovery likely involves some mechanism of learning, in which drug-associated classically or operantly conditioned behaviors must be dissolved or replaced. Both classically and operantly conditioned responses may undergo extinction, in which the conditioned stimulus (classical) or behavior (operant) is presented or performed in the absence of the unconditioned stimulus or reinforcement, respectively. A learned behavior is said to be “extinguished” when the conditioned stimulus no longer elicits a conditioned response (classical) or when the previously reinforced behavior is no longer performed (operant).120 It may be tempting to conclude from such observations that the learned behavior has been replaced or forgotten. However, learning theorists widely accept that extinction represents the weakening, not the abolishment, of learning.

Indeed, conditioned responses re-emerge in a variety of experimental situations.120 Spontaneous recovery refers to the return of a previously extinguished conditioned response following a time delay and in response to re-exposure to a conditioned stimulus. Reinstatement, renewal, and resurgence each refer to the return of a previously extinguished conditioned response following re-exposure to an unconditioned stimulus (reinstatement), changes in context (renewal), and the extinction of alternative behaviors (resurgence). These phenomena might be considered homologous to cue-induced craving (craving triggered by exposure to drug-associated cues) and relapse (a return to drug seeking), respectively.120–122 Thus, despite even long-term abstinence or intensive treatment, people with SUD still experience reactivity to drug-associated cues, craving, and relapse.123 Several translational models which were originally designed to assess the permanence of learned behavior have since been leveraged to investigate learning processes related or unique to substance use.121,124 Here, we will consider two of these models: renewal and resurgence.

The renewal effect refers to the context-dependent reemergence of extinguished conditioned behavior. Subsequent studies of renewal indicated that while initial conditioning or acquisition readily translated across context (ie fear conditioning learning in one setting could be elicited in another setting), extinction was at least somewhat context-dependent (ie extinction of fear conditioning in a therapeutic setting might not be effective at reducing fear responding in another setting).125,126 This effect had critical implications for treatments involved extinction or extinction-related processes, like Cue Exposure Therapy.127,128 In other words, strategies for maintaining abstinence developed in a therapeutic setting (via CBT, for example) might prove difficult to translate to the plethora of settings that comprise the “real” world, many of which have likely been previously associated with substance use.121 The presence of external positive reinforcement may aid in the prevention of such resurgence by providing external reinforcement of the exercise of such strategies. For example, the use of CM as an adjunctive practice in combination with CBT seems to improve treatment outcomes compared to CBT alone in some studies.42,54 Whether the strategy is derived from formal treatment or not, the maintenance of abstinence involves the employment of coping skills.129–132 Wherever these skills are developed, they must be employed in a variety of contexts during recovery. That CM reinforces the use of these skills by reinforcing abstinence may strengthen them.

Resurgence is a phenomenon during which an extinguished operant behavior (substance use) is recovered when the behavior that replaced it (ie reinforcement of abstinence via CM) has been extinguished. During CM, reinforcement maintains behavior that “replaces” substance use, thereby creating a positive contingency between that behavior and the reinforcer (abstinence is associated with availability of the reinforcer). However, as Bouton and Schepers133 suggest, the implementation of abstinence-contingent CM also creates a negative contingency between substance use and reinforcement (substance use is associated with absence of the reinforcer). In the traditional rodent model of resurgence, rats are first trained to press a lever reinforced by delivery of a food pellet. Then, a second lever is extended; the first lever becomes inactive (responses are recorded but not reinforced) and pressing the second lever is now reinforced by food pellet delivery, which ultimately eliminates responding on the first lever. As a final test of resurgence, both levers are extended but inactive (which will extinguish responding on the second lever). Generally, the rodent will return to responding on the first lever once the second lever is unreinforced. In the updated procedure, a negative contingency is introduced alongside the second lever such that the reinforcer delivery was contingent both upon the pressing of the second lever (positive contingency) and upon withholding responding on the first lever (negative contingency). In other words, to receive the food pellet, rodents had to both perform the new behavior (the second lever) and abstain from the old behavior (the first lever). Bouton and Schepers reported that the introduction of this negative contingency, analogous to an abstinence-contingency in CM, reduced resurgence, but did not completely abolish responding on the first lever.

One gap in the translation of Bouton and Schepers’ procedure to CM with humans is that, in the real world, the initial response (pressing the first lever) cannot be unreinforced; a return to drug seeking is likely to result in relapse. In other words, CM is a choice procedure, not an extinction procedure.134 In answer to this observation, an incentivized choice paradigm introduced another alteration to the resurgence procedures described above: responding on the initial lever remains reinforced throughout the procedure, though pressing of the second lever is reinforced by a larger reward. Although the increased incentive associated with the second lever was effective in reducing responding on the first lever, responding on the first lever resurged immediately following discontinuation of the reinforcing of the second lever.134 Taken together, these studies indicate that both the negative contingency and alternative source of reinforcement created by abstinence-contingent CM may be effective in reducing drug use, but only while the abstinence-contingency remains reinforced. Again, a likely mechanism of change in CM is the extinction and/or exchange of maladaptive behaviors, such that drug-seeking behavior is suppressed by alternative reinforcement and abstinence-maintaining behaviors are strengthened by external reinforcers.

The problem remains, at least in translational models, that the original (drug-seeking) behavior tends to resurge, despite evidence that CM can promote longer-term abstinence in people with SUD.6,43 Though translational models of behavior like those described above have contributed tremendously to learning theory, the study of behavior, and clinical research, all preclinical work involves the essential limitation (which is also a strength in experimental control) that rats or cats or pigeons or primates are not humans. This is certainly not to say that rats, cats, pigeons, and primates are incapable of sophisticated behavior, but rather that these creatures (to our knowledge) are not subject to the same cognitive, social, occupational, and cultural features that incentivize abstinence as humans. Thus, one critical component of SUD recovery that may not be fully addressed in translational models is the role of alternative or intrinsic motivation.

Intrinsic Motivation

Simply defined, intrinsic motivators are those which are derived internally, meaning without the influence of an outside force. The theoretical dimensions of intrinsic motivation include self-efficacy,135,136 curiosity,137,138 and competence, relatedness, and autonomy.139,140 Because it is an internal process, intrinsic motivation can be difficult to measure in active decision-making processes.141 On the other hand, extrinsic motivators are easier to control and observe and, therefore, far easier to assess.141

A common concern regarding the use of CM for the treatment of SUD involves the use of external reinforcement to promote behavior which it is assumed relies on intrinsic motivation (ie abstinence, recovery).72–75 For example, one might believe that without significant intrinsic motivation to change their behavior, someone with SUD will simply return to using a drug after CM-based reinforcers are no longer available. This concern, then, is rooted in a deeper concern that external rewards undermine intrinsic motivation, or the undermining effect. In 1971, Edward Deci divided participants into two groups, a control group who were not paid and an experimental group who were paid only on the second of three days of puzzle access.142 Participants in both groups had indicated interest in solving the puzzles before the experiment began. Deci observed that the participants in the experimental group spent significantly more time compared to the control group solving the puzzles during their break time on the second (paid) day and significantly less time on the third (unpaid) day. He interpreted this pattern as an indication that the availability of the extrinsic reward (payment) had degraded their initial intrinsic interest in the puzzle, evidenced by the lack of puzzling on the third day. The undermining or “overjustification effect” remains the subject of significant debate.139,141,143 Though some research continues to support Deci’s initial finding,144 other studies indicate that the use of extrinsic reinforcement in the form of verbal feedback or test scores can enhance intrinsic motivation.145,146

In recovery from SUD, a form of intrinsic motivation is assessed using the stages of change model, which prescribes treatment approaches based on an individual’s occupation of a particular stage denoting their theoretical level of motivation for change.147,148 In other words, these stages represent an individual’s desire to alter their current behavior. If extrinsic motivation in the form of CM undermines one’s intrinsic motivation for change, we might expect the latter to decrease during CM. In participants randomly assigned to receive standard treatment alone or with CM, pretreatment motivation conceptualized by the stages of change model did not predict treatment outcomes.149 The results of the same study indicated that assignment to the CM condition was not associated with change in motivation across the study, but was associated with longer duration of abstinence, suggesting that motivation may change as a function of abstinence, or successful maintenance of abstinence. The construct of motivation conceptualized within the stages of change is relatively specific to motivation to change. Though the motivation to change may play an important role in treatment readiness, the successful maintenance of abstinence also promotes self-efficacy, or an individual’s belief in their own ability to accomplish a task or produce a behavior successfully.136 Importantly, self-efficacy is distinct from motivation to change (ie, an individual’s motivation to produce a certain behavior), as well as other related constructs like autonomy (ie, an individual’s perception of freedom or control over their behavior) and treatment engagement (ie, an individual’s commitment to participating in treatment).

Overall, self-efficacy is protective against relapse during SUD recovery.150–154 For example, low self-efficacy ratings in ecological momentary assessment studies predict relapse by hours155,156 and pre-post improvements in self-efficacy are associated with better treatment outcomes.131,157 Taken together, these effects suggest that the maintenance of abstinence may promote self-efficacy, which in turn enhances long term treatment outcomes. To the extent that CM reinforces the development of abstinence-maintaining strategies, it may be that CM provides a structure in which people can develop self-efficacy through the employment of abstinence-maintaining strategies.

Balanced considerations of the undermining effect argue that the “detrimental” effect of extrinsic reinforcers on intrinsic motivation is likely situation specific. For example, extrinsic motivation in the form of legally mandated or otherwise compulsory treatment is generally not associated with better treatment outcomes.158,159 This discrepancy might be explained through the lens of perceived control or, indeed, self-efficacy. Let us here return to Tolman and MacFarlane. If in its provision of external reinforcement CM provides the external motivation for enacting the strategies required to maintain abstinence, then the resulting strategies are strengthened. Successfully employing these strategies may then in turn build belief in one’s ability to employ successful strategies in the future. Thus, extrinsic reinforcement may strengthen intrinsic reinforcement to the extent that it incentivizes standards of performance or mastery, rather than task completion and builds self-efficacy, which relies on an individual’s successful employment of learned abstinence-maintaining strategy.143

Decision-Making

When we engage in behavior change or replacement, the motivation for the new behavior is pitted against the motivation for the existing behavior. In the maintenance of abstinence, any intrinsic motivation for change or extrinsic incentive for abstinence, then, is pitted against the motivation for continued substance use. Numerous studies have consistently indicated that a variety of drugs commonly used by humans will function as reinforcers in preclinical studies using rodents and primates and lab studies involving human participants.118,119 These studies have contributed tremendously to scientific understanding of the neurobiological mechanisms and behavioral mechanisms of addictive drugs. However, the lack of choice between drug and non-drug reinforcers is a common criticism of exclusive access self-administration studies.160 SUD is not characterized simply by substance use, but by an array of behavioral symptoms that reflect a pattern of maladaptive substance use and substance-related decision making. For example, continued use of the substance despite known negative consequences (occupational, financial, social, etc.) features prominently in the diagnostic criteria for SUDs.12 In this way, recovery from SUD can involve the restoration of adaptive decision making and CM may offer an intermediate or transitional template for this change.

The use of alternative reinforcement to decrease drug-seeking has proven effective in rodent self-administering methamphetamine,161 cocaine,162,163 and nicotine.164 For example, rodents offered an exclusive choice between one lever positively reinforced by administration of intravenous cocaine and a second lever negatively reinforced by shock avoidance overwhelmingly directed their behavior towards the negatively reinforced lever, regardless of the dose of cocaine available.162 This preference for shock avoidance over cocaine reinforcement persisted when the shock intensity was decreased. Similarly, rodents will choose intense sweet tastes or palatable foods over drugs and the preference of for these non-drug reinforcers persists regardless of drug history and baseline availability of food.165–167 Importantly, both isomorphic (food vs food) and allomorphic (food vs drug) choices are sensitive to manipulation of the reinforcer probability (how certain the delivery of reinforcement is) and magnitude (how large the reinforcement is).168 Taken together, these findings broadly indicate that the choice of a non-drug over a drug is informed by the nature and quantity of the non-drug reinforcer. However, a choice in the real world is rarely, if ever, discrete or immediate. In models of exclusive choice, the non-drug reinforcer often supplants the drug reinforcer. For example, when rodents are forced to make an exclusive choice between social contact and drug infusion, the social reward suppresses responding for drug.169–172 However, rodents with non-exclusive, concurrent access to two independent levers reinforced by cocaine infusion or social interaction respond on both levers, demonstrating motivation for both cocaine and social interaction.173 Thus, the exclusivity of the choice is quite important. In abstinent-contingent CM, external reinforcers are mutually exclusive with substance use, reinforcing a discrete choice. Furthermore, reinforcers forgone by problematic substance use are often not immediate. The choice, then, is not between two immediately available reinforcers, but the short-term reinforcement of the drug versus long term social, occupational, financial, and health-related goals. The value of a reinforcer, and thus value-based decision making, is impacted not only by magnitude but also temporal duration.174–176 In general, immediate reinforcers are more valuable than delayed reinforcers of the same magnitude.177,178

One complicating factor in real world drug-associated decision making is that people with SUD have, by virtue of diagnosis, already engaged in a pattern of decision making that is considered disordered. Though now outdated models of addiction, namely the moral model, extol the role of individual choice in the pathology of addiction, modern theories of addiction consider the impact of chronic substance use on neural systems that fundamentally alter an individual’s capacity for decision making.3,15,17,179,180 Notably, the behavior involved in SUD is characterized by impulsivity and increased salience of drugs relative to other reinforcers. Impulsivity can be considered both facet of personality and a behavioral tendency, meaning that some people are more prone to impulsive behavior than others. As a behavioral tendency, impulsive behavior involves increased sensitivity to immediate reward relative to long term consequences and lack of response inhibition.

Interestingly, impulsivity is both a significant risk factor in substance use and substance use disorder181–188 and an outcome associated with drug use.189,190 Increased impulsivity can result in a tendency to prefer immediate reinforcers to longer-term, albeit larger, reinforcers (delay discounting).174 However, people with SUD also display greater attentiveness to drug relative to non-drug rewards and associated cues.191,192 The salience misattribution model for addictive-like behaviors postulates that chronic substance use dysregulates dopaminergic activity in the brain to such an extent that drug and drug-associated stimuli become “misattributed as highly salient”.193,194 This extensive salience accounts for the strong preference for drug relative to non-drug reinforcement, differential discounting of drug and non-drug reinforcers, and disordered motivational states like craving. In other words, decision-making is no longer properly regulated by reinforcer magnitude and probability, because drug reinforcers are faultily intensified. If decision making in SUD is impulsive and biased towards drug and drug-associated reinforcers, then recovery from SUD may involve restoration or correction of these processes.

Delay discounting is the tendency to choose smaller, immediate rewards over larger, delayed rewards.174 In other words, rewards that occur further away in time are, by virtue of being delayed, discounted relative to immediate rewards. Higher delay discounting, meaning an increased tendency to select the immediate reward, is a significant individual predictor of problematic substance use195 and, in general, people with SUD display greater delay discounting than those who do not. This effect is consistent in studies of people who use opioids,196,197 cocaine,198 tobacco,199–203 cannabis,204 and alcohol.56,181,182,205 There is also some evidence that the tendency to discount delayed reinforcers is increased during withdrawal. For example, participants maintained on buprenorphine undergoing opioid withdrawal displayed higher discounting of heroin.206 However, deficits in delay discounting are not a unique feature of SUD. For example, adults with hepatitis C are more likely to discount delayed rewards, independent of substance use history.207

It is also not clear whether recovery is associated with reductions in delay discounting. One study tracked changes in self-reported impulsivity and delay discounting in patients in an inpatient addiction medicine program.208 They report significant within subject change in impulsivity concomitant with positive psychiatric outcomes, but no change in delay discounting. However, Fontes and colleagues209 recently reported that individuals in sustained remission from SUD display decreased rates of delay discounting relative to individuals not currently in remission. Interestingly, individuals in remission from SUD also display a decrease in other maladaptive health-related behavior associated with delay discounting like poor eating and financial instability.210 Importantly, individual differences alone do not explain differences in delay discounting, which suggests that delay discounting may also develop based on exposure to environmental contingencies (eg the subjective value of the reinforcer).211 In other words, the tendency to delay discount when deciding between immediate drug and long term non-drug (eg maintaining one’s job, social acceptance, etc) reinforcers may also depend on the subjective value of the drug, which is increased for people with SUD.193 Taken together, these data suggest that delay discounting may present a unique challenge to individuals in SUD treatment and recovery due to relative immediacy of drug reinforcers relative to recovery-associated reinforcers and the increased value attributed to drugs. By providing an alternative immediate or intermediate reinforcer, CM may help to correct the imbalance in this decision making process.

Alongside clear evidence of an increased tendency to delay discount associated with SUD, it is possible that people with SUD experience a more general deficit in inhibitory control212 which also impacts response inhibition, a separate component of impulsivity. Response inhibition can be empirically assessed using the stop-signal or go/no-go tasks, which require participants to withhold pre-planned or automatic motor responses, like pressing a button when a cue appears on a screen. Deficits in the performance of both tasks have also been associated with substance use.213,214 Delay discounting can be ameliorated by precommitment215 or a decision-in-advance of the choice between the immediate and delayed rewards. However, it remains that the intention to forgo the immediate, smaller reinforcer for the larger, delayed one is not entirely synonymous with the ability to engage in the necessary inhibition to successfully resist the immediate (in the case of SUD, drug) reinforcer.216 This pattern is reminiscent of the lack of effect of pretreatment motivation on treatment outcome in some studies.149 Taken together, these data indicate that more is at play than simply the intention or motivation to alter future behavior. That SUD is associated with impulsivity, including both an increased tendency to delay discount and a decreased capacity for response inhibition, puts people with SUD at a disadvantage in pursuing recovery. As we will see, this altered capacity is underlined by neural changes in the brain that weaken executive function. In providing shorter-term, behavior-contingent, external reinforcers, CM offers an intermediate, exclusive choice during attempts to maintain abstinence which may serve as a template for long-term behavior change or as an opportunity to build self-efficacy through practicing inhibitory control strategies.

Neuroplasticity

The brain’s remarkable capacity for structural and functional reorganization in response to internal and external stimuli is called neuroplasticity. This allows the brain to adapt to experience, learning and storing information in the form of memory, but also enables the development of maladaptive behaviors. Drugs exert their effects on a neurochemical level through the manipulation of neurotransmission. The acute impact is the effect of the drug, but the chronic impact is a rewiring of the brain’s processing of reward.217 Though SUD may involve the use of different substances (eg opioids, nicotine) with specific neurobiological mechanisms (eg binding to opioid receptors, binding to acetylcholine receptors), all “addictive” drugs produce certain common neuroadaptations, particularly in the limbic system (involved in affective behaviors), the nucleus accumbens (NAc;involved in reward and motivation) and the prefrontal cortex (PFC; involved in executive function). In other words, SUD is associated with differences in reward processing and decision-making that can be linked to specific neuroplastic changes in reward- and decision- involved brain regions. The competing neural systems hypothesis218 suggests that choices for delayed outcomes require PFC (“executive system”) input and choices for immediate outcomes rely more heavily on the limbic system (“impulsive system”) which includes the NAc. During chronic substance use, the executive system is overwhelmed by the impulsive system, meaning that the PFC becomes inundated by limbic input.219–221 Though such neuroadaptations underlie the problematic behaviors associated with addiction that preclude recovery, the capacity for neuroplasticity may also be the key to recovery.100,217,222

The striatum, particularly the NAc, is involved in encoding the reinforcing value of drugs and drug associated cues.223–225 That SUD is associated with blunted striatal activation in response to non-drug cues and reinforcers is well established.226–229 One study evaluated striatal activation during acute tobacco abstinence supported by contingency-management.230 Participants completed a task in which they were told they might earn a tobacco or a monetary reward, while researchers collected functional magnetic resonance imaging (fMRI) scans of their brains. The imaging revealed increased activation in the striatum when participants were anticipating the tobacco reward and decreased activation when anticipating the monetary reward. Increased reactivity to drug relative to non-drug reward is indicative of increased motivation or anticipation of the drug reward, which biases decision making. In a preclinical model of voluntary abstinence, activation of the NAc increased, and inactivation decreased, relapse seeking.231 In clinical populations, striatal activation is also predictive of SUD treatment outcomes.230,232

The PFC, including the dorsolateral, ventromedial, orbitofrontal, and anterior cingulate cortices, is associated with executive functions like emotional regulation, working memory, and impulse control. The PFC exerts top-down control on lower-level circuitry, including the limbic system. SUD is associated with a weakening of this control, such that people with SUD display decreased activation in the PFC relative to healthy controls233 and decreased activation on an individual basis is a predictor of relapse.234 These alterations explain the aforementioned differences in extinction learning, intrinsic motivation, and decision-making. For example, inactivation of PFC and NAc promotes impulsive behavior.235 The PFC is also associated with self-efficacy, perceived control, and reduced impulsivity, all putative mechanisms of change during SUD recovery.236,237 Importantly, abstinence is associated with duration-dependent neural recovery, particularly in the PFC, which may reflect an increase or return of cortical and inhibitory control.100

A comprehensive review of the neurobiological changes involved in risk, pathology, and treatment of SUD is beyond the scope of the present review (we recommend Koob and Volkow 2010217 and Uhl, Koob, and Cable 2019).238 In general, the dyssynchronization of the PFC and the limbic system/striatum is implicated in a loss of top-down neural control that underlies impulsive choice and attentional and motivation bias towards drugs and drug-associated stimuli. Though the brain can recover during prolonged abstinence, these neural changes challenge SUD treatment. To the extent that CM promotes abstinence, provides immediate non-drug rewards, and allows for the development of strategies involving inhibitory control, it may facilitate the neural changes associated with SUD recovery, particularly the restoration of the PFC.

Conclusion

Despite decades of research, CM remains underutilized for the treatment of SUD relative to its efficacy in supporting abstinence.39–49 Barriers to the wider adoption of CM include negative attitudes from clinicians and the public, including that CM may undermine intrinsic motivation (ie the idea that people in recovery should want to recover)73–75 and that CM may only prove effective while extrinsic reinforcement continues.

First, we discussed the behaviorist roots of CM, particularly early work on operant conditioning and learning conducted by Thorndike, Skinner, Tolman, Macfarlane, and many others.33,34,84,85,91–93,239 From the early contributions of these scientists, we can infer much about the process of learning and the relationship between internal and external motivators. Rather than reject the existence of internal processes, B.F. Skinner questioned whether they could be measured empirically.86 Tolman and Honzik’s 1930 experiment demonstrated that latent learning might occur in the absence of extrinsic reinforcement (potentially maintained by intrinsic processes like curiosity) but that the presence of external reinforcers facilitates the demonstration of latent learning.91,92 In other words, external rewards direct, rather than enable, learning. Indeed, Macfarlane’s 1930 experiment assessing the flexibility of instrumentally conditioned behavior demonstrated that conditioning produced flexible goal-oriented behaviors rather than repeated motor patterns.93 Thus, if abstinence is the reinforcer-contingent goal, the strategies people develop during CM can likely still be employed to maintain abstinence in a variety of situations (ie whether extrinsic reward is available or not) and extrinsic reinforcement may direct the use of behavioral strategies learned in other settings (ie, other treatment settings, perhaps).

Next, we examined lab studies of post-extinction responding (ie, relapse), which reveal that reinforcer seeking can return after extinction as a function of time, context, and choice.120,122 The renewal effect suggests that behaviors learned in one context might not translate to other contexts.121,125,126 For example, tools developed in a therapeutic setting might not extend to the real world. If such tools are learned, perhaps they are latent, and CM could provide the external conditions that prompt performance. Similarly, resurgence reminds us that seemingly replaced behaviors can still reappear.121,133 The studies of resurgence both support the premise that CM can redirect substance use in the short-term by providing alternative reinforcers and call into question the long-term efficacy of CM for maintaining abstinence post-treatment. However, lab studies of resurgence cannot fully account for additional, perhaps intrinsic, motivation that further supports SUD recovery, like motivation to change. To the extent that CM can redirect drug use in the short-term, both the maintenance of abstinence and the practicing of abstinence-maintaining strategies.

Then, we discussed two constructs, self-efficacy and inhibitory control, which may be developed or strengthened during CM and endure when extrinsic reinforcement is discontinued. The undermining effect describes situations in which extrinsic reinforcers might degrade intrinsic motivation142,144 and is least likely to occur in situations where the extrinsic reinforcement structure is optional and contingent upon mastery or performance standards rather than engagement.143 CM employing mastery-contingent reinforcers might therefore promote self-efficacy by reinforcing the adoption of strategies that prove successful in maintaining abstinence. Low self-efficacy is a significant short-term predictor of relapse, so increasing self-efficacy might aid people in SUD recovery in maintaining abstinence.136,156,157 Additionally, decision-making that is biased towards a substance features prominently in the diagnostic criteria for SUD. SUD is associated with increased tendency to delay discount (eg to choose immediate drug reinforcer over long-term goals) and decreased response inhibition (engage in inhibitory control).56,195,196,198,199,204,212–214,216,219 Because CM can provide short-term reinforcers (eg relative to long-term health), it may support the maintenance of abstinence by reducing the need to make potentially discounted decisions. To the extent that CM reinforces the employment of abstinence maintaining strategies, it may also mitigate deficits in response inhibition. In our theory, the role of extrinsic motivation is to incentivize the performance of behavioral strategies for maintaining abstinence, exercising inhibitory control and bolstering self-efficacy upon success.

Finally, we considered the neurobiological changes associated with long-term substance use, including dysregulation of connectivity between the frontal lobes and the limbic system and the associated decrease in capacity for top-down inhibitory control.179,180,193,217,218,238 These changes underlie the symptomology of SUD by creating attentional and motivational bias towards a drug on a cellular level. Abstinence is associated with neural recovery, particularly in the frontal lobes, which mediate executive function.100 To the extent that these changes are involved in SUD recovery, CM might support or facilitate such modes of neural recovery by promoting abstinence, increasing self-efficacy, or supporting decision-making and continued abstinence may reciprocally reinforce the recovery of executive function.

In conclusion, we suggest that CM reinforces the performance of latent, goal-oriented, abstinence-maintaining strategies, which might promote self-efficacy, support healthy decision-making, and facilitate neural recovery, by providing a scaffold of external contingencies.

Future Research

Though significant research effort has been dedicated to demonstrating the efficacy of CM, future studies might incorporate the measurement of self-efficacy, delay discounting, and response inhibition across CM treatment alongside clinical outcomes to probe for correlation. Both the mechanism and the endurance of CM might be further elucidated by longitudinal studies which track these metrics beyond the conclusion of a treatment program. Finally, the theory of latent learning might be investigated in further systematic comparison between people in combined treatment (ie, CM and CBT, MI, or TSF) and individual treatment programs to isolate the impact of extrinsic reinforcement on skills developed in other treatment settings.

Author Contributions

Both authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.

Disclosure

The authors report no conflicts of interest in this work.