Behavioral therapies targeting reward mechanisms in substance use disorders

Abstract

Behavioral therapies are considered best practices in the treatment of substance use disorders (SUD) and used as first-line approaches for SUDs without FDA-approved pharmacotherapies. Decades of research on the neuroscience of drug reward and addiction have informed the development of current leading behavioral therapies that, while differing in focus and technique, have in common the overarching goal of shifting reward responding away from drug and toward natural non-drug rewards. This review begins by describing key neurobiological processes of reward in addiction, followed by a description of how various behavioral therapies address specific reward processes. Based on this review, a conceptual ‘map’ is crafted to pinpoint gaps and areas of overlap, serving as a guide for selecting and integrating behavioral therapies.

1. Introduction

The last few decades have been marked by an explosion in the number and type of behavioral therapies for SUD. In addition to well-established treatments like Cognitive Behavioral Therapy (CBT) and Contingency Management (CM), there is growing empirical support for newer-wave variants that focus on mindfulness- and future thinking, for example. This progress in behavioral therapy development corresponds to advances in our understanding of the neurobiology of addiction, particularly the role of reward processes. Many theories posit addiction as a disorder of reward, reflecting both pathological increases in the value assigned to drug rewards and cues, and decreases in the relative value of non-drug rewards (Koob, 2013a; Bickel et al., 2020; Garland, 2016; Koob and Volkow, 2010; Robinson and Berridge, 2008). Correspondingly, behavior therapies for SUD share common goals of shifting responding away from drug-related rewards and toward natural and healthier non-drug rewards. But within the broad concept of “reward”, neuroscience also now recognizes multiple partially independent reward processes, which may not always be influenced in parallel. Given the current armamentarium of behavioral therapies, it becomes possible to also consider how each engages (or not) with each of these specific reward processes.

To that end, we begin with a brief overview of key neurobiological processes in drug reward and addiction, followed by a description of current evidence-based therapies that target reward. Here we have noted a general, although not absolute, grouping based on the aims of the therapy, such that therapies tend to either primarily target reducing responsiveness to drug cues/rewards or primarily target enhancing responsiveness to natural rewards. We use this broad grouping for organizational purposes. The review then concludes by proposing a conceptual map or matrix, illustrating which reward processes appear to be targeted by each behavioral therapy, within each of these two broad aims (reduction of drug reward vs. enhancement of natural reward). This framework lays out a structured approach for developing and integrating behavioral interventions to produce better coverage of reward-related mechanisms in a clinically relevant fashion.

2. Reward: A promising target for the treatment of SUD

Neuroadaptation of the reward system is viewed as central to the development and maintenance of addition. The mesolimbic and mesocortical dopamine pathways are implicated in reward functioning and are sourced in the ventral tegmental area with projections to the nucleus accumbens and prefrontal cortex, respectively (Koob, 1992). The reward system is essential to adaptive functioning, aiding in the identification of, response to, and memory for stimuli needed for survival, such as food. Like naturally occurring rewards, all addictive drugs increase the concentration of dopamine in the nucleus accumbens, but they do so more strongly. With repeated use, a greater amount of the substance is needed to produce the same physiological and subjective effects, a process called tolerance. Oppositely, when the substance is taken away, withdrawal occurs, a process in which the brain and body attempt to return to the new homeostatic set point (Volkow et al., 2016).

In the allostatic model of addiction, chronic substance use acts as a constant stressor on the brain and body. The concept of allostasis entails an adaptation of the body’s homeostatic systems in response to external stressors (trauma, abuse, divorce) or internal stressors (infection, cancer, drug intoxication/use) in an effort to maintain stability. Allostatic “load” can be understood as the summative influence of chronic stressors over the lifetime, described by McEwen as “the price the body pays for being forced to adapt to adverse psychosocial or physical situations” (McEwen, 2000). Pervasive allostatic load, including chronic substance use, can result in pathophysiological change that shifts systems away from homeostasis to a state of dysfunction. In the case of substance use disorders, an allostatic shift manifests biologically as the dysregulation of neurotransmitter and hormone systems (e.g., dopamine in the reward circuitry pathway, corticotropin-releasing hormone in the HPA axis) and psychologically as the dysregulation of mood and alternations in reinforcer sensitivity (e.g., depression, anhedonia, hyperkatifeia) (Koob, 2021; Koob et al., 2014; Koob et al., 2020; Mcewen, 2001). Therefore, repeated use of drugs and alcohol leads to hedonic dysregulation (i.e., the emphasis on drug-seeking behavior at the expense of homeostatic goal-directed behavior). This neuroadaptation may contribute to the pronounced lack of interest in non-drug rewards and the “loss of self-control” observed in those with SUD (Koob and Volkow, 2016).

3. Translating neurobiology of reward to behavioral interventions

The understanding of different reward mechanisms can contribute to translating neuroscience research into the improvement of behavioral therapies for SUD. The neuroscience of reward suggests there are at least three major reward processes – “consummatory reward”, consisting of pleasure or “liking”; “motivational reward”, including motivation for/anticipation of reward or “wanting”; and “reward learning” which refers to changing subsequent behaviors based on a history of reward. These processes are partially separable at both the neural and functional levels (Webber et al., 2021; Zald and Treadway, 2017). Changes in all of these reward processes have been strongly implicated in SUD (Ray et al., 2023), but these reward processes may also be partially separable in terms of their association with SUD symptoms and behaviors.

For example, the incentive sensitization theory of addiction posits that cues that are repeatedly associated with administration of drugs and alcohol are imbued with incentive salience (Berridge and Robinson, 2003; Robinson and Berridge, 1993), capturing attention, activating affective states, and motivating drug-seeking behaviors. This theory distinguishes between the processes of anticipating or being motivated to seek a reward (“wanting”) and consuming (“liking”) the reward (Berridge and Robinson, 2016). Evidence shows that dopamine is more involved in the “wanting” aspect for reward and the response to drug cues, while the endogenous opioid system is more responsible for the “liking” aspect (Berridge et al., 2009). Critically, this theory posits that the effects of drugs on these two processes diverge in addiction. In animal models, motivational reward or “wanting”, defined by reward-seeking behaviors or drug consumption, occurs after repeated use in the absence of continued consummatory reward, or “liking”, as defined by facial expressions (Wyvell and Berridge, 2000). Clinically, this could explain why drug cues often elicit strong drug cravings and urges (which have explicitly been identified with “wanting” in this theory (Robinson and Berridge, 1993)), even in the absence of an intense pleasure response or “liking” for the drug after repeated drug use (e.g. (Arulkadacham et al., 2017; Grigutsch et al., 2019; Lambert et al., 2006)). This theory also accounts for how the dopaminergic effects of drugs may cause hyper-reactivity to drug cues that compete against reactivity to other non-drug reinforcers.

Regarding the third process of reward learning, it is well-known that dopamine neurons respond to the presentation of rewards. After repeated pairings with a conditioned stimulus, however, the dopamine neuronal firing shifts in time to the presentation of the conditioned stimulus (i.e., the reward predicting cue) rather than at the time of the reward itself (Schultz et al., 1997). If the conditioned stimulus is presented and the reward is withheld, dopamine neurons suppress firing below baseline at the time of the expected reward, a pattern known as reward prediction errors (Berridge et al., 2009). In SUD, reward prediction errors contribute to the development of cue-reactivity, or strong cravings in response to drug-related cues.

In addition to potentially accounting for different and partially separable aspects of addiction, different reward processes may also be differentially influenced by behavioral therapies for addiction. Below is a description of current reward-based behavioral interventions for SUDs, and a discussion of how research on these principles of reward may translate into targets for these therapies. Because mechanistic research has lagged significantly behind efficacy research in the field, these identifications must be considered tentative, but we hope this will spur further confirmatory research. As noted above, interventions generally are broken down into two broad categories by aim: therapies that aim to down-regulate or decrease drug cue reactivity and/or sensitivity to drug reward and therapies that aim to up-regulate or increase responsiveness to naturally occurring cues and rewards. Within each of these aims we can also seek to identify the role of specific reward processes (consummatory, motivational and reward learning) in contributing to this aim. This yields a 2 (decrease drug reward vs. increase natural reward) by 3 (consummatory, motivational, reward learning) matrix into which we can place therapeutic techniques, represented in Table 1, to which we will refer throughout.

Table 1.

Mapping therapeutic techniques in addiction onto the neuroscience of reward.

	Reducing Reward Value of Drugs			Increasing Value of Natural Rewards
Treatment	Consummatory - pleasure/subjective effects	Motivational - effort/anticipation	Reward learning - cue reactivity	Consummatory - pleasure/subjective effects	Motivational - effort/anticipation	Reward learning - cue reactivity
Cognitive-Behavioral Therapy		Cognitive restructuring	Functional analysis and coping skills training
Attentional Bias Modification Therapy			Increasing attentional control
Acceptance and Commitment Therapy		Acceptance and cognitive defusion	Functional analysis and optimizing psychological flexibility processes	Contact with present moment	Values and committed action
Contingency Management					Providing alternate rewards	Immediate reinforcement of non-use
Behavioral Activation					Increasing engagement in enjoyable activities
Episodic Future Thinking					Practicing anticipation of reward
Mindfulness-Oriented Recovery Enhancement		Metacognitive awareness of craving	Metacognitive awareness of triggers, functional analysis and coping skills training	Mindful savoring

4. Therapies that target reduction of drug-related reward

4.1. Cognitive behavioral therapy

Traditional cognitive behavioral therapy (CBT) is the most used and empirically supported therapy for the treatment of SUD. CBT originates from principles of classical conditioning and cognitive social learning theory (Bandura, 1986; Marlatt and Gordon, 1985). Through repeated stimulus-response associations, drug cues acquire the ability to elicit conditioned responses such as craving and drug-seeking behavior. The acquisition and maintenance of substance use behaviors also involves vicarious learning processes that take place by observing others’ behavior and the reinforcing consequences of the behavior. Cognitive processes mediate the effects of events or stimuli on behavioral outcomes. These processes include outcome expectancies about engaging in the behavior (e.g., a drink will reduce my stress) and self-efficacy expectancies (e.g., I cannot handle this situation without a drink), that serve to further reinforce substance use. Thus, the CBT model posits that substance use is a learned behavior and that the role of therapy is to help clients learn new, more adaptive behavioral and cognitive responses in situations associated with substance use.

The main goals of CBT are threefold. First, CBT focuses on developing a functional analysis of substance use by helping the individual identify drug-related cues, or “high-risk” situations that precede their substance use. This technique may target reward learning, particularly by bringing cue-related craving processes into conscious attention where they may be modified (see Table 1). Second, based on this analysis, CBT conducts cognitive restructuring – attempting to investigate and alter client perceptions of the short- and longer-term consequences that serve to maintain the behavior – this technique may target motivational reward by changing or devaluing anticipated effects of drug use (see Table 1). Third, also based on the functional analysis, CBT aims to identify and remediate deficits in coping skills by teaching new ways of responding effectively in high-risk situations without reliance on substances – this may assist with breaking learned cue-reward associations (see Table 1). Fourth, CBT incorporates relapse prevention skills training to prepare the client for relapse. These skills include lifestyle modifications aimed at reducing sources of stress and increasing pleasant and rewarding activities (this last reward-related technique is shared with Behavioral Activation and will be discussed more extensively in that section).

An extensive body of evidence from controlled studies has shown that CBT is effective in reducing craving, preventing relapse, and reducing substance use across a range of SUDs (Carroll, 1998; Dutra et al., 2008; Irvin et al., 1999; Magill et al., 2019; Magill and Ray, 2009). Meta-analytic reviews have been consistent in reporting effect sizes in the small-to-moderate range, depending on the contrast condition (Irvin et al., 1999; Magill and Ray, 2009). Most recently, Magill and colleagues (Magill et al., 2019) examined 30 randomized controlled trials that tested the efficacy of CBT for alcohol or other SUDs (cannabis, opioids, stimulants and poly-drug use) in contrast to minimal treatment (k = 5), non-specific therapy (k = 11), and specific therapy (k = 19). Pooled effect sizes for CBT when compared to no or minimal (treatment as usual) conditions were moderate and durable over follow-up. When CBT was compared to a non-specific or specific therapy modality, however, effects were small to non-significant. The lack of differential efficacy between specific therapies, known in psychotherapy research as the “Dodo Bird” effect (Marcus et al., 2014), calls into question the specific processes by which CBT works, arguing instead that non-specific factors common to all therapies lead to improved client outcomes, above and beyond any one intervention. When it comes to specific putative mechanisms of change in CBT for SUD, studies have found mixed evidence that increases in coping skills and self-efficacy mediate treatment effects (Kiluk et al., 2010; Witkiewitz et al., 2022; Witkiewitz et al., 2018). Beyond self-report, a handful of imaging studies have examined neurobiological mechanisms underlying CBT interventions in SUD (Zilverstand et al., 2016). In a study of smokers (Costello et al., 2010), those receiving 8 weeks of CBT treatment (vs. bupropion or placebo) showed a significant reduction in PET (Positron Emission Tomography) resting glucose metabolism in the posterior cingulate cortex, a brain region implicated in reward-related and attentional control processing (Kang et al., 2012; Leech and Sharp, 2014; Schacht et al., 2013). In an fMRI study of treatment-seeking patients reporting polydrug use (n = 12), 8 weeks of CBT treatment was associated with improved functional brain activity in regions implicated in cognitive control, response inhibition, motivation, and attention (DeVito et al., 2012). No such changes were found in the non-substance-using comparison group (n = 12). Taken together, this small but promising body of evidence suggests that treatment with CBT confers benefit by reducing reward sensitivity to drug-related cues; however, more well-controlled mechanistic research is needed.

4.2. Attentional bias modification therapy

Attentional bias modification (ABM) therapies focus on implicit processes involved in addictive behavior; specifically, the tendency to automatically direct attention toward substance-related cues. As described above, theoretical models of attention bias, stemming from classical conditioning, posit that substance-related cues acquire incentive-motivational properties that drive attention and behavior toward the goal of substance use (Christiansen et al., 2015; Marhe et al., 2013). Specifically, it is hypothesized that once drug stimuli become strongly associated with a drug’s reinforcing effects, those stimuli exert an attentional capture – they are more likely to be viewed (or become the focal point) from a complex visual field or array (Monem and Fillmore, 2017; Parvaz et al., 2021). For example, amidst a barrage of banner advertisements on a web site, an ad for alcohol is more likely to capture the attention of an individual with an alcohol use disorder relative to other competing but less salient ads. Similar models offer a bidirectional relationship such that substance-related cues increase subjective craving that in turn increases attention paid to the cues, and vice versa (Ryan, 2002).

The addiction Stroop task and visual probe task are commonly used in ABM interventions. In both tasks, attention bias is inferred by reaction time that is slower (Stroop interference) or faster (visual probe facilitation) when substance-related words or images are presented relative to neutral stimuli. Some but not all studies have shown that performance on these measures at treatment onset predicts subsequent relapse (see Christiansen (Christiansen et al., 2015) for review), with one ecological momentary assessment study showing attentional bias to peak just before relapse occurs (Marhe et al., 2013). The development of ABM as an intervention began with laboratory studies demonstrating the ability to train participants to look toward (attend) or away (avoid) from cues (e.g., (Field and Eastwood, 2005)), directly targeting reward learning and cue-reactivity processes (see Table 1). While these studies showed transient changes in attentional bias, evidence of generalization to novel stimuli or sustained effects on drug craving and/or drug self-administration outcomes has been relatively weak (see Clarke et al. (Clarke et al., 2014)).

Studies evaluating ABM training paradigms as a treatment intervention have yielded modest evidence of efficacy. Boffo and colleagues (Boffo et al., 2019) conducted a Bayesian meta-analysis consisting of 14 studies of interventions targeting substance-related cognitive biases for the treatment of alcohol and tobacco use disorders. Standard computerized paradigms involved exposing patients to substance- versus neutral stimuli, across multiple training sessions and training trials per session, with task contingencies arranged to reinforce increased cue reactivity toward non-substance-related cues. Most of these trials included a sham training comparator. Results indicated a small effect for ‘target engagement’ in terms of interventions producing reduction in attentional bias directly after the completion of the training. Intervention effects on measures of craving, reduction of substance use, and abstinence were weak and unreliable, indicating limited evidence to establish the clinical efficacy of ABM as a behavior change intervention in alcohol and tobacco use disorders. A study by Mayer and colleagues evaluated the efficacy of ABM in treatment-seeking individuals with cocaine use disorder using a controlled trial design comparing ABM therapy to a control therapy, with both conditions consisting of 5 training sessions of 2640 trials over 4 weeks (Mayer et al., 2016). Results showed a main effect of time only, with both groups exhibiting reduced attentional bias, craving and cocaine use at post-intervention. Interestingly, reaction times were slower for cocaine-related cues relative to neutral cues at baseline, reflecting an intentional shift in attention away from cocaine stimuli due to higher motivation or readiness to change in this treatment-seeking sample.

Since the Boffo et al. 2019 meta-analysis, recent ABM intervention studies have taken novel, multi-model approaches to advance interventions in this domain. These newer approaches have (1) utilized trial-by-trial regulation of attention bias via a cognitive reappraisal technique, integrating EEG and eye-tracking measures to mechanistically address and reduce sources of drug-related attention bias in people with cocaine use disorder (Parvaz et al., 2021); (2) employed transcranial direct current stimulation (tCDS) informed by EEG neuronal network topography to reduce attentional bias to methamphetamine cues in people with methamphetamine use disorder (Khajehpour et al., 2022); and (3) utilized cognitive bias modification training to reduce attentional bias to pain and opioid cues during medication visits for opioid use disorder pharmacotherapy (MacLean et al., 2023). Notably, these more recent ABM studies have been conducted in experimental settings. It is presently unknown if the results will generalize to changes in actual substance use rates in naturalistic settings, e.g., within the context of randomized clinical trials.

4.3. Acceptance and commitment therapy

ACT is a newer behavioral therapy comprising an empirical, principle-driven approach to distressing or unwanted internal psychological and physical experiences, common precipitants to substance use (Hayes and Levin, 2012). The emphasis is placed on the context and function of these experiences rather than their content as emphasized in traditional CBT. In third wave therapies, such as ACT, internal content (thoughts, feelings, and physical sensations) is not presumed to be causal, and thus the emphasis is not on changing the content of thoughts or related feelings. Rather, the context in which these internal experiences occur is altered in order to change the function of the unpleasant or distressing experiences (Hayes et al., 1999).

The functional contextual perspective is particularly relevant for SUDs and other addictive behaviors. Unpleasant and unwanted internal thoughts and feelings are ubiquitous, especially among individuals struggling with alcohol or drug use. Common feelings include sadness and anxiety; thoughts such as, “I’m an addict and always will be,” or “I need a drink,” as well as physical sensations associated with alcohol and/or drug withdrawal. Perhaps more salient in addiction is that these internal experiences are both precipitants and consequences. Substances are used to both eliminate/control unpleasant thoughts and feelings as well as to manage the physical and psychological consequences resulting from excessive substance use, resulting in a “revolving door” of experiential avoidance (Hayes and Levin, 2012), or repeated attempts to change the form, frequency, and/or intensity of these distressing internal experiences, even when ineffective and in the face of negative consequences. A repeated reliance on experiential avoidance via substance use paradoxically exacerbates the exact experiences individuals are trying to avoid (Hayes et al., 1999).

In ACT, the overarching construct or guiding process is psychological flexibility, i.e., patterns of behavior regulated by the six, interrelated ACT processes involved in either expanding or narrowing behavioral repertoires (Hayes et al., 2013). The six processes include 1) acceptance, as opposed to avoidance, of distressing internal and external events; 2) cognitive defusion, a disconnection from distressing thoughts that often become connected with self and identity; 3) contact with the present moment, non-judgmental awareness and connection with internal and external experiences; 4) self-as-context, or perspective taking, viewing oneself within a context of experiences (internal and external); 5) values, guiding beliefs as to what is important; 6) committed action, behaviors directed by identified values. Identifying and amplifying values and engaging in committed actions toward values arguably focus more on increasing responding to natural rewards, similar to therapies described in Section 3, below. As such, ACT stands out as a therapy involving techniques that may address both aims (reduction of drug reward and enhancement of natural reward) and multiple types of reward processes (see Table 1) – potentially reducing motivation and expectancies for drugs via acceptance and cognitive defusion, impacting reward learning through shared techniques with CBT, such as functional analysis, and increasing consummatory and motivational reward for natural rewards via present-moment orientation and values-congruent committed actions.

From 2010 to 2020, at least a dozen meta-analyses or reviews of acceptance and mindfulness-based approaches have included addiction studies (Bautista et al., 2019; Garland and Howard, 2018a; Goldberg et al., 2018; Grant et al., 2017; Ii et al., 2019; Jiménez, 2012; Lee et al., 2015; Öst, 2014; Stotts and Northrup, 2015; Sancho et al., 2018; Byrne et al., 2019; Gloster et al., 2020). A large meta-analysis of 60 ACT studies conducted by Öst in 2014 reported a small effect size across all comparisons (Hedge’s g: 0.42) and included five RCTs of ACT for SUD (two in opioid use disorder, one in methamphetamine use disorder, and two that accepted people with various use disorders). Öst concluded that ACT was “possibly efficacious” for SUDs (Öst, 2014), which was similar to efficacy determinations reached by other reviews (Lee et al., 2015; Stotts and Northrup, 2015). Stotts and Northrup reported RR/OR effect sizes for ACT in SUDs (three studies in opioid use disorder, one in methamphetamine use disorder and three in mixed substance use disorders) ranging from small-to-large, depending on the assessment timing (i.e., end-of-treatment [EOT] or follow-up periods) and comparison group (passive vs. active control conditions) (Stotts and Northrup, 2015). Similarly, Lee and colleagues reported small-to-moderate effect sizes (g = 0.45, 95 % CI = 0.15, 0.74, z = 2.95, p = 0.003, k = 5) favoring ACT at post-treatment for drug-use outcomes (including the same use disorders noted above, and adding five studies examining smoking cessation) (Lee et al., 2015). Also in 2015, A-Tjak and colleagues published a meta-analysis that included 8 SUD studies (503 participants including two studies of opioid use disorder, two studies of smoking, one of methamphetamine use disorder, one of alcohol use disorder, one with mixed use disorders and one of a behavioral addiction [pornography]) with results favoring ACT (Hedges’ g = 0.40, SE = 0.13, 95 % CI: 0.15–0.66, p = 0.002) (A-tjak et al., 2015). Recent reviews by Byrne and colleagues included third wave studies addressing alcohol use disorders (AUD) and indicated promise for ACT and other mindfulness approaches in treating AUD, noting a need for research with first-line comparison conditions (Byrne et al., 2019). Several studies have supported the effectiveness of ACT for smoking cessation (Bricker, 2023), including a large trial of a smoking cessation application (iCanQuit) that resulted in a 1.7 times increased likelihood of achieving 7-day point prevalence abstinence at 6 months relative to a guideline-based application developed by the National Cancer Institute (Bricker et al., 2014). In 2019, Ii and colleagues adopted a mechanism-based approach and broadly analyzed studies of treatments that targeted psychological flexibility. These psychological flexibility-based third wave treatments were associated with almost a 10 % increase in substance discontinuation compared to first-line interventions (e.g., brief MI, 12-step groups), i.e., 33.6 % vs. 24.8 %, respectively, across 5 studies of mixed substance use disorders, three studies in opioid use disorder, and one in alcohol use disorder (Ii et al., 2019).

5. Therapies that target increased response to natural rewards

5.1. Contingency management

Contingency management (CM) is an evidence-based behavioral intervention that provides reward incentives (e.g., cash, gift cards) for achieving a clinically-relevant target behavior, such as abstinence in the treatment of SUD. CM is based on operant conditioning theory that utilizes the concept of positive reinforcement to motivate behavior change. In behavioral economic terms (Bickel et al., 2014), rewards provided by CM engage decision-making processes related to delay discounting, i.e., the tendency for individuals with SUDs to exhibit relatively shortened temporal horizons in which delayed consequences have significantly less impact on their current behavior; and drug demand, i.e., skewed valuation for drug rewards compared to non-drug rewards. CM addresses both by providing non-drug rewards that are immediate (reduce temporal discounting) and that compete with the reinforcing value of drug (reduce demand). In doing this we posit that CM may act on both motivational reward for non-drug rewards, and reward learning for non-drug rewards via the provision of more immediate rewards for abstinence (see Table 1).

Although the parameters of a specific CM intervention may widely vary, most interventions share a number of common characteristics. Objective verification of the target behavior, especially of drug abstinence, is critical, as reliance on self-report may be inaccurate and result in inadvertently rewarding drug use. In regards to the incentive delivery schedule, previous research supports two key components to optimize CM efficacy: 1) an escalating schedule in which the incentive value increases with consecutive negative samples, as opposed to a fixed incentive value (Roll et al., 1996); and 2) a response cost in which no incentive is delivered following drug use, and subsequent incentive values following drug abstinence are “reset” to the lowest incentive value (Roll and Higgins, 2000). While not always implemented, CM may also incorporate increased response requirements over the course of treatment in which either the required number of days abstinent is increased or frequency of incentive delivery is decreased.

Numerous studies and meta-analyses have repeatedly demonstrated the efficacy of CM. For example, a recent systematic literature review of 69 CM studies from 2009 to 2014 observed an average effect size (Cohen’s d) of 0.62 (95 % CI: 0.54, 0.70), ranging from moderate to large treatment effects across a variety of substance use disorders (nicotine, alcohol, cocaine, cannabis, methamphetamine, opioids and polysubstance use) (Davis et al., 2016). Despite positive findings, critics of CM have questioned the long-term effects of CM once incentives are discontinued. While this argument can be made for other behavioral and pharmacological treatments for SUD, there is empirical evidence supporting CMs long-term efficacy. A recent meta-analysis including 23 randomized trials (15 in stimulant use disorder, 7 in polysubstance use and one in opioid use disorder) that utilized CM for SUDs with biochemically confirmed abstinence up to 1-year follow-up yielded results showing CM to be superior to comparison treatments (e.g., CBT, 12-step, etc.) with an overall weighted average OR effect size of 1.22 (95 % CI: 1.01, 1.44) for abstinence at a median of 24 weeks after CM ended (Ginley et al., 2021). Cost has been identified as another barrier to CM adoption. One variation of CM is the prize or fish bowl method, based on an intermittent schedule of reinforcement in which drawings for the chance to win a prize are delivered upon occurrence of the target behavior (Petry et al., 2000). Prizes range in value from $0 (“good job”) to small, medium, and large- (e.g., $100) sized rewards. Prize bowl CM is equally efficacious to the traditional voucher CM in promoting abstinence (Petry et al., 2005), but lower in cost and less complicated to administer. That CM “works” mechanistically by competing with the reinforcing value of drug was suggested in a recent study showing a significant decrease in cocaine demand over time among individuals receiving 4 weeks of high-magnitude abstinence-based CM (Yoon et al., 2021). In summary, CM is a highly effective and flexible treatment for addressing SUDs. While challenges to CM implementation do exist, various procedural and technological advances have increased the utility and availability of CM outside of clinical research settings.

5.2. Behavioral activation

Behavioral activation (BA) is a brief, structured psychotherapeutic approach based on reinforcement theory, which suggests that substance use develops and is maintained when there is a lack of competing, non-substance-related rewarding activities (Vuchinich and Tucker, 1988). The theory of BA is rooted in research related to the treatment of depressive disorders, which has focused heavily on cognitive therapies and CBT. Although CBT primarily focuses on cognitive strategies geared toward changing negative thoughts and attitudes, it also includes strategies to change behavior patterns (Beck, 1979). The original cognitive therapy manual for depression included strategies intended to activate people within their environments. These structured behavioral strategies (Beck, 1979; Jacobson et al., 1996) formed the basis for BA as a stand-alone evidence-based treatment for depressive disorders (Ekers et al., 2014), and subsequently adapted for the treatment of SUD with promising results (Fazzino et al., 2019).

Multiple sources of reinforcement have been identified which contribute to the initiation and maintenance of substance use (Koob and Moal, 1997). The intoxicating effects of various substances positively reinforce continued use, whereas the removal of aversive stimuli experienced through withdrawal from substances is negatively reinforced when substance use is resumed (Koob, 2013b). Therefore, when an individual enters BA treatment with a goal of abstinence, sessions focus on addressing the reinforcement deficit by increasing engagement in goal-directed and enjoyable activities (Fazzino et al., 2019). BA includes a wide range of behavioral strategies, including but not limited to monitoring daily activities, scheduling activities, generating a list of value-driven behaviors, intentionally engaging in substance-free behaviors, and assessing pleasure and mastery related to engaging in the identified behaviors (Daughters et al., 2008; Daughters et al., 2018; Lejuez et al., 2011). The intention of these behavioral strategies is to increase access to and engagement in substance-free activities. It is believed that by increasing access to alternative reinforcers, substance use will decrease (Reynolds et al., 2011). We posit that in neuroscientific terms BA represents an attempt to engage the motivational reward system for non-drug rewards (see Table 1).

At least two systematic reviews have examined the efficacy of BA in the treatment of co-occurring depression and substance use, yielding mixed results. The recent meta-analysis by Pott and colleagues did not provide evidence of differential effectiveness between BA and controls on depression outcomes (k = 5; N = 195; SMD: −0.10, CI −0.51 to 0.30; p = 0.62) or substance use outcomes (k = 4; N = 151; SMD: 0.17, CI −0.34 to 0.69; p = 0.51) at follow up, potentially due to limitations in the number and quality of studies included in their review (two in smoking, three in polysubstance use) (Pott et al., 2022). There was evidence supporting the acceptability of BA interventions over control conditions in terms of higher average session attendance rates. Martinez-Vispo et al. conducted a narrative systematic review of the effects of BA on substance use and depression (Martínez-Vispo et al., 2018). Of the 6 RCTs included (two in smoking, two in polysubstance use and one in opioid use disorder), 2 reported significantly higher abstinence rates with BA compared to the control conditions (Daughters et al., 2018; MacPherson et al., 2010). A third pilot study of BA for smoking cessation reported a greater number of days to first lapse after discharge for the BA versus the control group, with other outcomes favoring BA but not significant (Busch et al., 2017). A fourth study assessing the feasibility of BA for depression symptoms in patients receiving drug and alcohol treatment found a 17 % increase in days abstinent in the BA group, but this was not statistically significant (Delgadillo et al., 2015). The authors noted that the systematic review indicates the potential for BA to improve substance use in the context of comorbid depression, but that more research with larger samples is needed (Martínez-Vispo et al., 2018).

Only one systematic review to date has examined the efficacy of BA to address substance use without targeting co-occurring depression. Fazzino et al. in 2019 reviewed reinforcement-based interventions, including 7 BA and three behavioral economics (BE) trials targeting substance use outcomes and/or mechanisms (Fazzino et al., 2019). Of the BA studies, three targeted nicotine, one targeted alcohol, two targeted any drug or alcohol, and one targeted methamphetamine use. Most of the studies (6/7; 86 %) reported favorable substance use outcomes. The review included 5 RCTs, with 4 (80 %) reporting significant differences in outcomes between the BA and controls, including higher odds of abstinence in three of the trials and greater treatment retention in one trial (Daughters et al., 2018; MacPherson et al., 2010; Magidson et al., 2011). Three of the RCTs reported moderate to large effects (ORs: 2.2 to 3.6) and one reported small to moderate effect sizes (Ors: 1.12 to 1.82). Only one randomized trial which targeted alcohol use among college freshmen did not find significant changes in alcohol consumption, although it did find significant reductions in hazardous drinking with a moderate effect size (OR = 2.2; d = 0.44) (Reynolds et al., 2011). The systematic review also reported results of two single-arm trials with both reporting significant reductions in substance use compared to baseline (Banducci et al., 2013; Mimiaga et al., 2012). Overall, the current body of evidence supports the efficacy of BA for improving substance use outcomes. More rigorous studies are needed to determine the distinct benefits of BA over other evidence-based therapies and to test the mechanisms of action driving BA efficacy.

5.3. Episodic future thinking

Episodic Future Thinking (EFT) is a relatively recent intervention based on reinforcer pathology theory, supported by principles of behavior economics that target underlying sub-optimal decision-making processes. Whereas CM, described above, enhances the availability of immediate non-drug alternative rewards, the goal of EFT is to increase the saliency of future positive activities and rewards. EFT is supported by an extensive body of research linking behavioral economic mechanisms of delay discounting and drug demand to change in drug use. Specifically, EFT targets impulsive decision-making patterns associated with addiction and typically characterized by excessive preference for smaller, more immediate, and overvalued rewards (e.g., drug use) relative to larger, delayed ones (e.g., better health, financial stability). Both greater delay discounting and drug demand are associated with various aspects of drug use severity and relapse (Bickel et al., 2014; Amlung et al., 2017; Strickland et al., 2020; Zvorsky et al., 2019).

EFT involves training individuals to identify and vividly experience personalized, rewarding events at various timepoints in the future. Additional support is provided by asking individuals to elaborate on these events using who, what, where, when, and how questions (e.g., “who is there with you?”, “what are you doing?”, etc.). For each event, individuals are asked to create a distinct cue that can be subsequently delivered through various outlets such as text messages to prompt retrieval in real-world decision-making situations. While the underlying neurobiological mechanisms related to EFT’s impact on decision-making are yet to be elucidated, EFT presumably improves decision-making processes related to both delay discounting and drug demand by increasing saliency and influence of future rewards on current behavior. We have classified this as representing an attempt to strengthen the motivational reward system for non-drug rewards by engaging in an explicit practice of anticipating future reward (see Table 1).

Given the infancy of the EFT field, too few studies have been conducted to perform meta-analysis, however pilot studies to date have generally supported this approach. When compared to a control condition such as episodic recent thinking (ERT) protocols, evidence suggests that EFT decreases delay discounting, drug demand, and actual drug use. In the case of alcohol, an initial study demonstrated that EFT compared ERT significantly decreased delay discounting and demand for alcohol among non-treatment seeking individuals with alcohol use disorder (N = 50) (Snider et al., 2016). In a subsequent study among treatment-seeking individuals with alcohol use disorder (N = 52), initial training followed by two weeks of twice-daily cues resulted in significantly reduced alcohol consumption in the EFT versus ERT group (Athamneh et al., 2022). Additionally, changes in delay discounting from the pre- to post-intervention assessments significantly predicted alcohol consumption. Similar findings have been observed supporting the efficacy of EFT in studies of alcohol (Chang and Ladd, 2023; Patel and Amlung, 2020; Voss et al., 2022) as well as other drugs such as tobacco, cannabis, and opioids (Athamneh et al., 2021; Craft et al., 2020; Sofis et al., 2022; Garcia-Perez et al., 2022; Ruhi-Williams et al., 2022). Given the relative ease and low resources required for implementation, EFT shows great promise based on current findings as either a stand-alone or additional supportive intervention for SUDs.

5.4. Mindfulness-oriented recovery enhancement

Mindfulness Oriented Recovery Enhancement (MORE) is a promising group-based treatment for addiction and substance misuse that integrates existing cognitive-behavioral and mindfulness approaches to addiction while adding an explicit focus on learning to savor natural rewards (Garland, 2013). MORE has elements in common with CBT, including functional analysis of triggers, coping skills and reappraisal. It also has aspects in common with ACT, such as utilizing acceptance, present-focused mindfulness, and an emphasis on the guiding value of a meaningful life. MORE could be considered a sub-type of mindfulness-based relapse prevention approaches more broadly (Li et al., 2017). We have already discussed links between these techniques and motivational/reward learning processes in the prior sections (see Table 1). However, we are covering MORE as our example of mindfulness-based techniques, because it incorporates an additional, more explicit focus on using mindfulness to increase the hedonic value of non-drug rewards in addition to disrupting automatic, learned valuations of drug rewards (Garland, 2016). This represents an explicit attempt to strengthen the functioning of the consummatory reward system for non-drug rewards (see Table 1).

The critical techniques in MORE are mindfulness, reappraisal, and savoring (Garland, 2016). These techniques are applied iteratively across 8–10 2 h sessions (Garland, 2013). Each session typically contains an instruction on applying mindfulness to an addiction-relevant topic, an in-session mindfulness practice, debriefing on the practice, and homework. The initial sessions contain foundational instructions in mindfulness, with the goal of having patients build greater cognitive control that can be applied in later sessions. Additional sessions use this foundational technique to introduce mindful reappraisal and mindful savoring. These same techniques are used in subsequent sessions to address triggers, craving, coping, suppression/acceptance, interdependence/relationships and relapse prevention (Garland, 2013).

MORE is a newer technique, with the first pilot study in alcohol use disorder being published in 2010 (Garland et al., 2010). Studies of MORE have been heterogenous, both in conditions addressed (alcohol use disorder, long-term opioid use for chronic pain both with and without opioid misuse, opioid use disorder with chronic pain, internet gaming disorder, and comorbid substance use/psychiatric disorders) and outcomes (Parisi et al., 2022a). A 2022 multi-level meta-analysis summarized 8 clinical trials using MORE for SUD (Parisi et al., 2022a), finding a small to moderate effect on craving (standardized mean change = −0.42), which included effect sizes for opioid craving, general substance craving, alcohol craving and video game craving. In this same meta-analysis, MORE demonstrated a moderate effect size (standardized mean change = −0.54) on “addictive behaviors”. This included effect sizes for opioid misuse (on a combination of self-report, interview and urine drug screen measures), self-reported problematic internet use and internet gaming, and self-reported drug use. Since publication of that meta-analysis, two additional studies have evaluated substance-relevant outcomes of MORE delivered by telehealth. One examined MORE for comorbid opioid use disorder and chronic pain, showing that MORE reduced illicit drug use (assessed via a combination of EMA and saliva testing) (Cooperman et al., 2024). Another tested MORE plus a just-in-time intervention for stress triggered by a wearable heart monitor for opioid-treated chronic pain patients, finding MORE reduced opioid cravings (Garland et al., 2023a). Studies of the mechanisms of MORE have generally shown positive predicted effects on subjective, behavioral and neural measures of responsiveness to non-drug rewards, and in some cases indicate these effects mediate clinical improvements in craving and pain (Garland et al., 2023b; Garland et al., 2014; Garland et al., 2015; Garland et al., 2019; Garland and Howard, 2018b). MORE also appears to effectively enhance meditative processes and use of reappraisal (Garland et al., 2010; Parisi et al., 2022b). Finally, MORE leads to improvements in heart rate variability, a parasympathetically influenced measure generally thought to index emotional flexibility and resilience (Garland et al., 2010; Garland et al., 2014). Overall, MORE stands out for largely successful efforts to confirm the mechanistic basis of this treatment.

6. Mapping therapeutic techniques onto the neuroscience of reward

As noted throughout, behavioral therapies targeting reward processes in addiction generally have the objective of either decreasing the pathological reward value of drugs or increasing the value of non-drug rewards. However, also as noted above, the neuroscience of reward suggests there are at least three reward processes – pleasure or “liking”; motivation/anticipation or “wanting”; and learning from reward – which are partially separable at both the neural and functional levels and may be targeted to different degrees by different therapies. Combining these two major objectives with the three reward systems produces the potential “map” showing which reward processes are targeted by behavioral therapies that aim to downregulate drug rewards or upregulate natural rewards that we have attempted to build throughout this review. As summarized in Table 1, this conceptual mapping highlights both gaps and areas of overlap in addressing reward processes. This map also points to potential principled or systematic ways to combine therapies based on their relative coverage of different reward processes, as described in the following section.

6.1. Multipronged therapy approaches to act on multiple aspects of reward

Examining the “map” above in Table 1 suggests that current therapies generally do not address all aspects of reward that are relevant to addiction, and suggests theoretical rationales for combining particular therapies or therapeutic techniques that act on complementary aspects of reward. For example, although CM is among the most consistently effective addiction therapies that addresses reward, estimates across studies suggest it is still only successful in a minority of patients (Dutra et al., 2008). CM provides alternate rewards, but the value of those natural rewards may also be impaired in the pathology of addiction. Table 1 would suggest also targeting the consummatory pleasure or “liking” system to enhance the value of natural reward. Thus, one particularly effective combination might be to add either ACT or MORE to CM. These therapies share a number of techniques, and also address reward functioning across several domains that are not captured by CM. As one specific example, MORE uses mindful savoring to enhance responses to natural rewards, while CM can provide the opportunity to experience more alternate rewards. This combination might be particularly important to explore for stimulant use disorders, where CM is currently the only effective behavioral treatment (Bentzley et al., 2021). Indeed, in a recent study by Schmitz and colleagues (Schmitz et al., 2024), ACT significantly increased response to an abstinence-based high-magnitude CM intervention for cocaine cessation, suggesting potential synergy in combining therapies that target these different reward processes.

Table 1 also underscores the observation that none of the behavioral therapies directly target the downregulation of the immediate subjective effect of drugs – a strategy that has proven successful for medication therapies. Indeed, this scientific premise has formed the basis for numerous trials evaluating behavioral-pharmacological combinations over the years (Dakwar and Nunes, 2016; Carroll et al., 2004; Schmitz et al., 2008; Tardelli et al., 2018). For example, the add-on benefit of CBT with naltrexone pharmacotherapy for alcohol use disorder is thought to reflect synergistic effects of coping skills to reduce craving along with opioid medication to block reinforcing effects associated with alcohol consumption. Newer candidate medications, such as glucagon-like peptide-1 receptor agonists (GLP-1RAs) appear to have similar utility in reducing drug reinforcing effects, potentially complimenting the effectiveness of behavioral therapies (Eren-Yazicioglu et al., 2020).

In addition to combining behavioral therapies with each other or with medications, combining behavioral therapies with non-invasive brain stimulation is a promising avenue to pursue for targeting reward in SUD (Wesley and Lile, 2023). Brain stimulation produces long-term potentiation effects and when performed prior to or during psychotherapy, can facilitate learning that occurs during a session (Isserles et al., 2011; Luber et al., 2007). The effects of brain stimulation are also state-dependent (Silvanto and Pascual-Leone, 2008). Thus, engaging in psychotherapy or a behavioral task prior to brain stimulation can also modulate the effects of brain stimulation treatment. In the mood and anxiety-related disorders field, brain stimulation with CBT has shown promising results (Isserles et al., 2011; Neacsiu et al., 2018). This idea is now reaching the SUD field. For example, Dinur-Klein et al. (Dinur-Klein et al., 2014) found that participants exposed to a smoking cue-reactivity task prior to brain stimulation (deep transcranial magnetic stimulation, TMS) significantly reduced cigarette smoking compared to sham stimulation. Potentially, TMS and other stimulation techniques could more precisely and effectively target the specific neurocognitive mechanisms of reward laid out in Table 1. Future studies should consider the combination of brain stimulation therapies with specific behavioral therapies having complementary mechanisms of action.

6.2. Challenges and opportunities in translating neurobiology to behavior therapy implementation

While this paper highlights strides made in applying knowledge about neurobiological mechanisms of reward into behavioral therapies, there are inherent challenges in the translational process, a topic that has been extensively covered in the literature (Field and Kersbergen, 2020; Goldstein and Volkow, 2011; Vandaele and Daeppen, 2022; Venniro et al., 2020). Translation is limited not only by differing neuroanatomy, but also differing contexts and study settings that cannot be as well-controlled in human research compared to animal research. Further, as Ray and Grodin discuss, preclinical and clinical models are often not well aligned even concerning their definition of addiction (Ray and Grodin, 2021). For example, the DSM-5 characterizes SUDs as mild, moderate, or severe and encompasses a broad range of use, whereas it is common in preclinical models to expose the animals to higher drug quantities resulting in greater withdrawal symptoms and drug-seeking behaviors. Several DSM-5 criteria additionally rely on patient self-report responses and not just observable behavior, such as “taking the substance in larger amounts or for longer periods of time than intended”, which provides further challenges for translation.

Despite these challenges, some studies have attempted to identify neurobiological changes in response to treatment to test putative mechanisms. In principle, an effective neurobiological-based behavioral therapy would reverse dysregulated reward processes in the brain (Berridge and Robinson, 2016). While evidence to date does not demonstrate ‘reversal’ per se, studies have indicated changes in specific neural regions in response to behavioral treatments (Mestre-Bach and Potenza, 2023). Studies of CBT that have incorporated pre- and post-treatment fMRI scans have found evidence supporting the hypothesis of improved cognitive control-related neural processing as a mechanism associated with treatment efficacy (e.g., (DeVito et al., 2012; Brewer et al., 2008; DeVito et al., 2017; DeVito et al., 2019; Yip et al., 2014; Kober et al., 2014)). Mindfulness-based therapies have been shown to increase resting state functional connectivity, consistent with the hypothesis of improved attentional control as the target mechanism of change (Zilverstand et al., 2016). Wiers and colleagues (Wiers et al., 2015) observed reductions in neural alcohol cue reactivity in the medial prefrontal cortex in patients receiving cognitive bias modification training compared to the control group, suggesting that ABM therapies may have an effect on reducing the motivational salience of drug cues. Further studies incorporating neuroimaging measures are needed to test hypothesized mechanisms underlying response to various behavioral therapies (Mestre-Bach and Potenza, 2023).

In the absence of neuroimaging, behavioral measures have been utilized to indirectly assess mechanisms through which a specific therapy is believed to induce changes in brain systems. As mentioned in Section 5.4, the work by Witkiewitz and colleagues serves as an exemplar in this regard. Mindfulness is conceptualized as reducing self-reported craving through two candidate neural pathways: “top-down” modulation of executive function and “bottom-up” reactivity to cues. Accordingly, measures of inhibitory control, motivation, stress- and cue-reactivity have been used to investigate putative brain-behavior mechanisms of change resulting from mindfulness practice (Witkiewitz et al., 2013). In addition to mindfulness, other cognitive therapies such as CBT and ACT are thought to have similar effects on brain structure and function, making it challenging to ascertain therapy-specific neurobiological mechanisms. While this paper focuses on reward mechanisms, it is important to acknowledge that behavioral therapies also target non-reward mechanisms. Coping skills training in CBT, for example, aims to improve responding to drug cues, but also “works” by altering outcome expectancies and boosting self-efficacy. Behavioral activation enhances engagement in enjoyable activities, while simultaneously improving mood. Thus, the centrality of reward processes in the efficacy of behavioral therapies remains uncertain.

Finally, establishing neurobiological reward mechanisms of change in behavioral therapies may not always generalize to clinical trials. Even when replicated in clinical trials, the effectiveness of such therapies may not translate to real-world community settings where less-than-desired treatment effects are often observed. This translational gap between efficacy and effectiveness has been extensively reviewed and is beyond the scope of this paper. Of relevance here is the importance of understanding the key underlying mechanisms of effect within an intervention. This understanding makes it possible to adapt the treatment to better fit the target population, without compromising the core mechanisms that produce desired outcomes.

7. Conclusions

In conclusion, although reward-related processes in SUD are targeted by many current behavioral therapies, and the neuroscience of reward has informed the development of many of these therapies, there are still gaps in our arsenal of treatments. By systematically examining these therapies from the perspective of the neuroscience of reward – particularly the existence of multiple reward processes – and examining which therapies and techniques target response to drug rewards versus natural non-drug rewards, we have constructed a conceptual map of potential targets. This map can be used to guide the development of combined or integrated treatments for SUD that more systematically address the multiple aspects of reward implicated in SUD.

Data availability

No data was used for the research described in the article.