Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2022 Aug 1.
Published in final edited form as: Drug Alcohol Depend. 2021 Jun 6;225:108795. doi: 10.1016/j.drugalcdep.2021.108795

Reinforcer Pathology in Cocaine Use Disorder: Temporal Window Determines Cocaine Valuation

Sarah E Snider 1, Jamie K Turner 1, Samuel M McClure 2, Warren K Bickel 1
PMCID: PMC8282732  NIHMSID: NIHMS1711148  PMID: 34119880

Abstract

Aims:

The Experimental Medicine Approach offers a unique perspective to determine clinical behavior change by engaging a target underlying the cause of a disorder. The present work engaged a novel target of addiction, Reinforcer Pathology, in two studies to test changes in behavior among individuals with cocaine use disorder.

Methods:

In Study 1, n=44 participants engaged the temporal window with episodic future thinking (EFT), a positive prospection exercise. Changes in temporal view and cocaine valuation were tested using delay discounting and behavioral economic demand, respectively. Additionally, a computational model assessed the relative reliance on the near- and far-sighted systems during EFT. In Study 2, n=71 engaged the temporal window with a negatively-valenced hurricane scenario to test the opposite effects on window length and cocaine valuation.

Results:

Results demonstrated systematic and symmetrical engagement of the behavioral target. Study 1 robustly replicated previous work, wherein EFT lengthened the temporal window and decreased cocaine valuation. Moreover, EFT increased the weighting of the modeled far-sighted system, increasing the relative impact of long-term discounting decisions. Study 2 produced opposite outcomes, shortened temporal window and increased cocaine valuation.

Conclusions:

This approximately equal and opposite reaction to the manipulations supports reinforcer pathology theory and implicates the temporal window over which rewards are valued as a target to be pushed and pulled to produce clinically meaningful behavior change. Using the Experimental Medicine Approach as a guide, future work should identify new potential interventions to engage reinforcer pathology and use the clinically relevant outcomes as a litmus test for mechanism.

Keywords: Reinforcer Pathology, Experimental Medicine Approach, Episodic Future Thinking, Delay Discounting, Behavioral Economic Demand, Cocaine Use Disorder

1. Introduction

Cocaine use disorder persists as an enormous problem in the United States. No approved medications exist and psychosocial interventions are only moderately successful. Moreover, drug abuse continues to be diagnosed and treated based on a cluster of symptoms (i.e., DSM criteria) rather than mechanistically. We propose instead to tackle diagnosis and treatment strategies inductively (Bickel et al., 2019). The Experimental Medicine Approach, originally proposed by Claude Bernard and supported by the NIH, incorporates four steps to determine the clinical behavior change as a function of target engagement of an underlying cause of the disorder (Bernard, 1957; Riddle and Science of Behavior Change Working Group, 2015).

Step 1. Identify a Target:

From our work and others, we have identified an important and viable target for cocaine use disorder called Reinforcer Pathology. Reinforcer Pathology is an emerging framework that provides a conceptual understanding of why drugs are overvalued in substance use disorders (Bickel et al., 2014a, 2011a). Central to this framework is the idea that the relative reinforcing value of short-term (e.g., drugs) and long-term (e.g., employment or relationships) rewards are integrated over an individual’s temporal window (Bickel et al., 2019). For example, brief, intense, immediate, and reliable reinforcers such as cocaine offer enormous reinforcing value in the short-term but over longer time frames produce cumulative negative consequences. In contrast, long-term reinforcers such as employment or relationships offer more variable and lower intensity reinforcement in the short-term, but over time have cumulative positive consequences. The reinforcer pathology theory posits that an individual’s ability to evaluate delayed consequences at the time of choice, whether future outcomes are negative or positive, influences decision utility and predicts an individual’s choice tendencies.

Step 2. Develop an Assay:

The temporal window, the window over which individuals integrate the value of rewards, can be measured with delay discounting, which measures the reduction in the value of a reinforcer as a function of its delay (discount rate). Decades of literature demonstrate that excessive discounting of the future is ubiquitous in substance use disorders (Amlung et al., 2016; MacKillop et al., 2011). Moreover, delay discounting serves as a behavioral marker at all stages of addiction and predicts therapeutic outcomes (Bickel et al., 2014b). Yet, reinforcer pathology is the only contemporary theory of addiction that identifies delay discounting as a determinant of addiction.

Step 3. Engage the Target:

Recent work has demonstrated that narrative scenarios, whether self- or experimenter-generated, can engage the length of one’s temporal window. In particular, episodic future thinking (EFT), in which participants self-generate and vividly describe positive future events at different time points in the future (Atance and O’Neill, 2001), decreases discount rates (i.e., lengthens the temporal window) in alcohol drinkers (Snider et al., 2016a), cigarette smokers (Stein et al., 2018, 2016), and obese individuals (Daniel et al., 2013a; Stein et al., 2017). We and others have demonstrated the reverse effect by engaging the target with a negative scenario provided to participants. In particular, negative scenarios describing sudden income shock or a devastating hurricane increases discount rates (shortens the temporal window) in smokers and obese individuals (Bickel et al., 2016b; Mellis et al., 2018; Snider et al., 2019).

Step 4. Test Behavior Change:

Target engagement clarifies the value of the Experimental Medicine Approach; that is, if target engagement impacts other aspects of the disorder, then the target is a central mechanism within the disorder. Alternatively, if target engagement fails to change other aspects of the disorder beyond the target, then the target may not be a central generator of the disorder. Such identification could set the stage for the development of novel interventions for substance abuse. Previous studies engaging the temporal window with both EFT and negative narrative scenarios have shown changes in the subjective value and actual consumption of the participant’s commodity of choice in alcohol, cigarettes, and food (Daniel et al., 2013b; O’Neill et al., 2016; Snider et al., 2019, 2016b; Stein et al., 2016). Previous research has yet to show the impact of increases and decreases in the temporal window in the same research report, nor have they investigated those effects in cocaine use disorder.

We employed these foundations to mechanistically test the reinforcer pathology theory in two studies. Study 1 engaged the length of the temporal window target using EFT and measured its effects on the valuation of cocaine. In addition, we analyzed the effect of EFT through the lens of the competing neurobehavioral decision systems model (CNDS) (Bickel et al., 2016a, 2007; McClure and Bickel, 2014). This model is based on neuroscience findings suggesting that the temporal window is produced from two processes that have distinct impacts on behavior. A myopic brain system has a high discount rate and preferentially values near-term rewards, and a more far-sighted brain system has a low discount rate and therefore more equitably values delayed rewards. We used a computational instantiation of this model that estimates the discount rate for each system through fitting discounting over near-term and distant time scales. We were able to test between three means by which the temporal window may be expanded: the myopic system may be less myopic (its discount rate may decrease), the far-sighted system may be more far-sighted (decreased discount rate), or the relative reliance on the far-sighted system compared with the myopic system may be increased (changes in a weighting of the far-sighted system in determining behavior overall). In Study 2, we engaged the temporal window to shorten the time over which rewards are integrated with a negatively valenced hurricane scenario and measured its impact on the value of cocaine, in addition to testing the generality of a shortened assay of the target. The symmetrical but opposite changes in the temporal window confirmed the relevance of reinforcer pathology to cocaine dependence and demonstrated support for delay discounting as a mechanism that may guide therapeutic interventions to come.

2. Study 1

2.1. Methods

2.1.1. Participants

A total of n=44 participants who met DSM-5 criteria for cocaine use disorder completed a 2–3h session. All procedures for this study were reviewed and approved by the Virginia Tech Institutional Review Board. See Supplemental Material for details on recruitment practices and eligibility criteria.

2.1.2. Procedure

Following informed consent, participants completed a battery of assessments including a Beck Depression Inventory (BDI)(Beck et al., 1996), and Beck Anxiety Index (BAI)(Beck et al., 1988). Participants also completed baseline assessments of the Positive and Negative Affect Scale(Watson et al., 1988)(PANAS) and cocaine craving (modified from smoking urges questionnaire(Tiffany and Drobes, 1991). See Supplemental Material for details on each assessment.

The study had two key-dependent measures. First, delay discounting measured an individual’s temporal window. Two iterations of the $100 adjusting amount delay discounting task(Du et al., 2002) were presented, Money-Money and Cocaine-Money. The Money-Money task titrated a small amount of money now vs. $100 after some delay and the Cocaine-Money task titrated a small amount of cocaine now vs. $100 after some delay. Both tasks measure the target, the temporal window. The Money-Money task is the more common laboratory measure, but the Cocaine-Money task is more ecologically valid (Bickel et al., 2011b). Second, the cocaine purchase task (Bruner and Johnson, 2014), a measure of cocaine value, asked participants to make hypothetical purchases in grams of their preferred form of cocaine at an assortment of different price points.

Participants were then asked to generate a total of seven vivid future (episodic future thinking; EFT) or, the control condition, recent past (episodic recent thinking; ERT) cues in an interview-style session. Details of the cue generation procedure were previously described in Snider et al. (2016b). In short, research staff guided EFT participants through the process of generating realistic, positive future cues at the following time points: 1 day, 1 week, 1 month, 3 months, 1 year, 5 years, and 25 years. ERT participants generated cues of positive events that occurred in the recent past at the following timepoints: yesterday night from 7pm-10pm, yesterday evening from 4pm-7pm, yesterday afternoon from 1pm-4pm, yesterday midmorning from 10am-12pm, yesterday morning from 7am-10am, the night before last night 7pm-10pm, the evening before last from 4pm-7pm. Participants were prompted by the research staff via a survey script to provide supporting details for each cue (e.g., “Where will you be?”, “Who will you be with?”, “What will you be hearing, tasting or smelling?”).

Following the cue generation, participants were asked to repeat the PANAS, cocaine craving, two delay discounting tasks, and cocaine purchase task to assess EFT-induced changes. At baseline, no cues were presented; however, following cue generation, EFT or ERT cues were presented with each discounting and purchase task decision.

2.1.3. Data Analysis

Discounting data was screened for systematic responses based on Johnson and Bickel (2008) criteria. In brief, violation of Criterion 1 is defined as a >20% increase in indifference point compared to the preceding indifference point. The present study only excluded individuals who violated Criterion 1 more than once (i.e., jumping). Violation of Criterion 2 was defined as <10% difference of indifference points between the first and last delay (i.e., lack of discounting).

A total of 5 participants (n=3 EFT; n=2 ERT) were excluded from the money-money analyses and 6 participants (n=4 EFT; n=2 ERT) were excluded from the cocaine-money analyses for violating Criterion 1. In addition, a total of 5 (n=3 EFT; n=2 ERT) participants were excluded from the money-money analyses for violating Criterion 2. Each of these exclusions were unique participants with the exception of one ERT participant who was excluded from both tasks. If a participant was excluded from money-money tasks analyses, their data was retained for cocaine-money analyses and vice versa. A total of 13 (n=8 EFT; n=5 ERT) participants violated Criterion 2 in the cocaine-money task; however, we retained those individuals because the task required a choice between immediately available cocaine and delayed money. A plausible set of responses included a situation in which a participant devalued cocaine entirely, requiring responding for the delayed money choice even at the longest delays. Moreover, a Fischer’s exact test indicated no systematic difference in Criterion 2 violations by group.

Delay discounting rates were analyzed using a modification to traditional area under the curve analysis, AUClogd, as suggested by Borges et al (2016). This method provides a theory-free measure of valuation across delays without disproportionally overvaluing the longest delays. The AUClogd values were then standardized by dividing them by the longest logged delay to produce a range from 0 to 1. Subsequent analyses use the CNDS model (see below) to identify putative mechanisms underlying changes in discounting across conditions.

Cocaine demand data were screened for systematic responses based on Stein et al. criteria (Stein et al., 2015). In brief, purchasing that violated the Trend (i.e., increases in purchasing as price increases) or Bounce (i.e., greater than 25% increase in purchasing compared to that at the lowest price) criterion were excluded from analyses. Overconsumption was defined as purchases of greater than 50 grams of cocaine/crack for self-use over 24 hours, a quantity 10× greater than the maximum tolerated dose for a regular cocaine user (Claustre et al., 1993). We note that no Reversals (i.e., resuming purchasing at higher prices after zero consumption at a lower price; breakpoint) were present in the current dataset. In addition, given that the demand analyses were performed on group data, zero purchasing at all prices (n=1 EFT group) was retained as it could conceivably be a valid data point. A total of four participants were excluded from demand analyses. Two participants (n=1 EFT; n=1 ERT) violated one or both of the Trend and Bounce criteria. The other two participants (n=1 EFT; n=1 ERT) were excluded for Overconsumption criteria violations. Group demand for cocaine was derived using the exponentiated demand equation (Koffarnus et al., 2015).

Competing Neurobehavioral Decision Systems: Model Fitting

We expected that our EFT manipulation would influence delay discounting by increasing attention to future time points. Models of delay discounting have been proposed that explicitly separate present bias and future thinking. One form of these models posits dual-valuation systems that differ in the rate of discounting (McClure et al., 2007; van den Bos et al., 2014). According to this model, subjective value (V) is determined by the weighted (ω) sum of these two systems:

V(r,t)=r[(1−ω)βt+ωδt]

In this equation, r is the undiscounted value of the reward available at delay t and β and δ are the discount rates for the two systems (0 < β < δ ≤ 1). This model closely approximates the hyperbolic discount function (see Study 2), so that its explanatory value does not lie in the quality of its fit per se, especially since it has three free parameters (β, δ, ω versus of just k). Instead, the model promises to help understand why discounting changes. β and δ capture the impact of each valuation system on behavior, with smaller values indicative of reduced valuation of delayed rewards. If β or δ are influenced by EFT, then it suggests that the nature of the underlying valuation processes is affected. The final parameter, ω, estimates the degree to which each system contributes to behavior overall. If ω changes then it suggests that the relative contribution and importance of the two systems are affected, but that the underlying processes remain fixed. An analogy with cocaine use may be that the near-term (i.e., drug high, β) and long-term (i.e., negative personal and professional effects, δ) consequences of drug use remain consistent, but the importance of the different consequences in determining behavior changes (ω).

The β-δ model was fit to the indifference points generated in the delay discounting task by minimizing the sum-squared error between the model and empirical values:

argminβ1δ1ω∑i[D(ri,ti)−V(ri,ti)]2

where D(ri, ti) are the indifference points measured at each future time point ti. Best fitting model parameters were found using the minimize function in Python’s SciPy package. Parameters were constrained so that β > 0, δ > β, δ ≤ 1, and 0 ≤ ω ≤ 1.

Confidence intervals for model parameters and best fitting V were calculated by bootstrapping to avoid assumption of normally distributed residuals. We generated 10,000 samples by randomly drawing N datasets from our data with replacement, where N varied for the EFT and ERT groups to match the number of datasets retained after excluding data as discussed above. For each sample, we estimated best-fitting model parameters and fitted value function. We then identified the 2.5% smallest and largest values for each measure and generated 95% confidence intervals that spanned this range.

2.2. Results

2.2.1. Demographics

Participant demographics are presented in Table 1 for the money-money and cocaine-money tasks. No differences in demographics emerged between groups. No differences in positive or negative PANAS affect scores or cocaine craving scores emerged between groups before or after the intervention.

Table 1.

Demographic characteristics of participants randomly assigned to either the EFT or ERT Control group for the money-money and cocaine-tasks. Gender and Race variables are percentages based on the group. All other variables are group means (±SEM).

Money-Money Task
N Gender (% male) Race (% caucasian) Age (SEM) Monthly Income (SEM) Years of Education (SEM)
Active EFT (16) 56.30% 37.50% 48.0 (1.5) 685.90 (199.3) 12.1 (0.2)
Control ERT (18) 72.20% 37.50% 41.4 (2.7) 971.1 (223.8) 12.5 (0.3)
p-value 0.48 >0.99 0.052 0.35 0.39
Cocaine-Money Task
N Gender (% male) Race (% caucasian) Age (SEM) Monthly Income (SEM) Years of Education (SEM)
Active EFT (18) 66.60% 27.80% 47.1 (1.8) 673.3 (181.2) 12.3 (0.2)
Control ERT (20) 65.00% 30.00% 42.1 (2.7) 942.8 (215.3) 12.5 (0.3)
p-value >0.99 >0.99 0.14 0.35 0.64
Open in a new tab

2.2.2. Delay Discounting

2.2.2.1. Money-Money Task

Figure 1c compares money-money delay discounting AUClogd between EFT and ERT groups at baseline and post intervention. A 2-way ANOVA (between groups, within subjects on timepoint) revealed a significant interaction, F(1,32) = 7.22, p=0.011, indicating that EFT had a significantly greater impact on delay discounting than did ERT. The main effects of timepoint (Baseline vs. Post), F(1,32) = 2.65, p = 0.11 and group difference, F(1,37) = 0.44, p=0.51 were not significant.2.2.2.2.

Figure 1.

Figure 1.

Open in a new tab

Between Groups Money-Money Discounting. (A; Left) Group mean indifference points (±SEM) at baseline. (B; Center) Group mean indifference points (±SEM) post episodic thinking. (C; Right) Group mean standardized area under the curve (AUClogd) for both groups (±SEM).

Cocaine-Money Task

Figure 2c compares cocaine-money delay discounting between EFT and ERT groups at baseline and post the intervention. A 2-way ANOVA (between groups, within subjects on timepoint) demonstrated a significant main effect of timepoint (Baseline vs. Post), F(1,36) = 7.14, p=0.011 and a significant interaction, F(1,36)= 7.52, p = 0.0094. Sidak’s multiple comparison’s post hoc tests produced trend differences between EFT and ERT groups post intervention (p = 0.07), but not at baseline, (p=0.34).

Figure 2.

Figure 2.

Open in a new tab

Between Group Cocaine-Money Discounting. (A; Left) Group mean indifference points (±SEM) at baseline. (B; Center) Group mean indifference points (±SEM) post episodic thinking. (C; Right) Group mean standardized area under the curve (AUClogd) for both groups (±SEM).

2.2.2.3. Competing neurobehavioral decision systems model

To better understand the mechanisms that underlie the changes in the discount rate, we fit a model based on hypothesized dual processes used in evaluating the future (McClure and Bickel, 2014). This model separates discounting into discounting over the near term (β) and discounting over the long-term (δ), with a weighting factor (ω) that is larger when the long-term discounting process has a greater impact on subjective value. We fit the discount functions in each of the conditions, producing good fits for all conditions (data within 95% confidence bounds at all time points and R2 values greater than 0.83 for all conditions; Fig. 3A, B, D, E). Please note that best-fitting values for δ ranged from 0.997 to 0.999 across the task conditions (see Supplementary Figure 1). These values correspond to discounting by 30% to 66% over one year (0.999365=0.694, or approximately 30%, and 0.997365=0.334, or roughly 66%).

Figure 3.

Figure 3.

Open in a new tab

Competing neurobehavioral decision systems (β-δ) model fits. Fits were generated for all subjects individually with model parameters calculated pre-manipulation (Base) and differences in parameter values after manipulation (Δ). (A, B) Model fits were within standard errors (shaded regions) in EFT and ERT in the money-money conditions. (C) Fitted parameter values are plotted with 95% confidence intervals for before. Changes in parameter values (Δ for β, δ, ω) in which confidence intervals do not include zero are considered significant. (D, E, F) Model fits and parameter values for the cocaine-money conditions.

Model parameters were compared within-participant by examining values before the intervention and the difference in parameter value due to the intervention (shown as change, Δ, for pre-post intervention in the figure). The most consistent effect across conditions is that the weighting factor (ω) increases after EFT but not after ERT (bar plots of parameter estimates show bootstrapped 95% confidence intervals; values for Δ that are above zero indicate a significant change in the associated parameter). Specifically, EFT increased ω in both the money-money and the cocaine-money discounting tasks whereas ERT did not change ω in either of these measures. The critical test is whether the change in weighting factor (Δω) differs between EFT and ERT, and this was not observed. These results should be considered to only reveal plausible mechanisms underlying the change in discounting observed in the main analyses, above.

The estimated values for the long-term discount rate (δ) did not change across any of the manipulations. However, estimates for (β) decreased in the money-money but not the cocaine-money condition for both ERT and EFT. This effect is evident in the raw data. The indifference value at short delays (1 day, in particular) is greater at baseline than after EFT and ERT. We do not believe that this change in β is meaningful for two reasons: (1) β decreases equally in EFT and ERT so that the difference in EFT versus ERT is not significant, and (2) β is larger at baseline in the money-money conditions (~0.8) than in all other conditions (ranging from 0.4–0.6 in cocaine-money pre- and post-manipulation, and 0.4–0.6 in money-money post-manipulation).

Overall, these model fits suggest that EFT does not change the processes that are used to evaluate the future but rather changes how these processes are used when evaluating a future reward. Specifically, the model parameters that summarize the evaluation mechanisms in the dual processes (β, δ) are not changed in EFT relative to ERT. This indicates that the rates of discounting captured by the two parameters are not affected by the manipulation. However, the relative importance of the two processes, represented by the weighting factor ω, for aggregate subjective value is impacted by EFT. The relative importance of the long-term evaluation process is approximately doubled by EFT and is unchanged by ERT (baseline ω is roughly equal to the change in ω following EFT). So, while the components of the discounting process are unaltered, the discount rate change observed in the analyses of AUClogd may indicate greater reliance on the process with a lower discount rate (δ) and less reliance on the process with a higher discount rate (β).

2.2.3. Demand

Figure 4 illustrates averaged group demand curves of both EFT and ERT at baseline (EFT R2 = 0.90; ERT R2 = 0.84) and post episodic thinking (EFT R2 = 0.86; ERT R2 = 0.89). Following cue generation, intensity of demand for cocaine was significantly lower in the EFT group, compared to ERT, F(1,468) = 32.8, p<0.0001 (Fig. 4B). No difference in elasticity between the groups was present. In addition, no difference was present at baseline between groups in intensity or elasticity (Fig. 4A).

Figure 4.

Figure 4.

Open in a new tab

Hypothetical purchasing of grams of cocaine between EFT and ERT groups. (A; Left) Group mean gram purchasing (±SEM) at baseline. (B; Right). Group mean gram purchasing (±SEM) post episodic thinking. Curve fits based on exponentiated demand equation (Koffarnus et al., 2015).

3. Study 2

In order to demonstrate a robust push-pull engagement of the reinforcer pathology target using narrative, we conducted a second study aimed to increase discount rates and demand for cocaine using a negatively valenced hurricane scenario. In this study, we used a shortened version of the delay discounting task (Koffarnus and Bickel, 2014) in order to determine generality.

3.1. Methods

Participants (n= 71) were recruited from Roanoke, VA and the surrounding areas. The inclusion criteria for the study were the same as in Study 1. In this abbreviated study design, participants were asked to complete baseline BDI and BAI questionnaires in addition to baseline money-money delay discounting and cocaine purchase tasks. The delay discounting task was slightly different than in Study 1 in that participants completed a $1000 5-trial adjusting delay discounting task procedure (Koffarnus and Bickel, 2014). In short, the first choice presented the option of receiving $500 now or $1000 in 3 weeks. The participant’s response to this question adjusted the time delay in the next trial, while the dollar amount of both rewards remains the same. The participants’ choice on the final trial determined their ED50, and subsequently their ln(k) value by assuming that discounting follows a hyperbolic discount function (Koffarnus and Bickel, 2014).

Individuals were then asked to read and imagine themselves in a devastating hurricane scenario or the control condition, a mild storm scenario, as described in Snider, et. al.(2019) and the Supplemental Material. Participants were randomly assigned to the hurricane (H) (n=37) and minor storm (MS) (n=34) groups. Participants were instructed to read along as a research staff member read the scenario out loud and to take the next 15 seconds to think about the scenario and how it would make them feel before moving on.

Following the scenario consideration, participants were then asked to repeat delay discounting and cocaine purchase tasks in order to assess scenario-induced changes.

Data Analysis

A total of 6 participants were excluded from demand analyses. Five participants (n=2 hurricane; n=3 mild storm) violated the Bounce criterion, and one participant violated both the Trend and Bounce criteria. In addition, eighteen participants (n=13 hurricane; n=6 mild storm) were excluded for Overconsumption criterion violation. A Fischer’s exact test indicated a non-significant trend between-group exclusion (p=0.11), however, inclusion or exclusion of Overconsumption violators demonstrated the same effect in demand between the hurricane and mild storm group.

3.2. Results

3.2.1. Demographics

Participant demographics are presented in Table 2. No differences in demographics emerged between groups, with the exception of age. That is, the hurricane scenario group was significantly older than the mild storm group on average (p=0.048). However, differences between group ln(k) (see below) remained significant when controlling for age as a covariate.

Table 2.

Demographic characteristics of participants randomly assigned to either the Hurricane or Mild Storm group. Gender and Race variables are percentages based on the group. All other variables are group means (±SEM).

N Gender (% male) Race (% caucasian) Age (SEM) Monthly Income (SEM) Years of Education (SEM)
Hurricane (37) 64.90% 35.10% 47.5 (1.8) 842.8 (172.8) 12.1 (0.2)
Mild Storm (34) 41.10% 44.10% 42.2 (2.0) 489.9 (124.6) 12.2 (0.3)
p-value 0.06 0.47 0.048* 0.11 0.74
Open in a new tab
*

p<0.05.

3.2.2. Delay Discounting

Figure 5 demonstrates a significant difference between obtained ln(k) values between the hurricane and mild storm group t(69) = 2.224, p=0.0294, indicating more discounting in the participants who engaged with the hurricane scenario.

Figure 5.

Figure 5.

Open in a new tab

Between Group Hurricane and Mild Storm Discounting. Group mean ln(k) (±SEM). *p<0.05.

3.2.3. Demand

Group demand curves between the hurricane (R2 = 0.75) and mild storm (R2= 0.81) groups demonstrated that the intensity of demand for cocaine was significantly greater in the hurricane scenario group, compared to the mild storm, F(1,548) = 35.79, p<0.0001 (Fig. 6). No difference in elasticity was present between groups.

Figure 6.

Figure 6.

Open in a new tab

Hypothetical purchasing of grams of cocaine between Hurricane and Mild Storm groups. Group mean gram purchasing (±SEM). Curve fit based on exponentiated demand equation (Koffarnus et al., 2015).

4. Discussion

The present research demonstrated systematic and symmetrical engagement of the behavioral target Reinforcer Pathology in individuals with cocaine use disorder. First, Study 1 robustly replicated the effect of EFT observed in other populations, this time demonstrating that target engagement to lengthen an individual’s temporal window over which rewards are integrated, decreased discount rates. In parallel, EFT decreased valuation of cocaine in cocaine-dependent individuals. Moreover, using model-fitting we demonstrated that EFT engagement does not directly affect the competing decision systems, but rather increases the weighting of δ, to increase the relative impact of long-term discounting decisions. Second, Study 2 produced opposite outcomes (increased discount rates and valuation of cocaine) by engaging the target with a narrative intervention to shorten the temporal window. This approximately equal and opposite reaction supports reinforcer pathology theory and implicates the temporal horizon over which rewards are integrated as a target to be pushed and pulled, suggesting that it may be used to produce clinically meaningful behavior change. Together, these data explicate the value of the Experimental Medicine Approach which we discuss in the following paragraphs.

Reinforcer Pathology describes a state in which an individual’s excessive discounting of the future may increase his or her valuation for immediate reinforcers and decrease it for delayed alternative reinforcers (Bickel et al., 2019). The present studies corroborate previous evidence (Daniel et al., 2013a, 2013b; O’Neill et al., 2016; Rung et al., 2019; Snider et al., 2019, 2016a; Stein et al., 2016) that delay discounting may be intervened on to produce correlated changes in the valuation of immediate cocaine in individuals with cocaine use disorder. Moving cocaine valuation in both directions by different manipulations that target delay discounting supports the role of the temporal horizon as a novel determinant of drug value. Although the present study did not measure real world cocaine use, these data may provide the preliminary evidence and context to utilize the experimental medicine approach to better understand these relationships, and engage the appropriate target. Future work should aim to demonstrate those next steps.

In addition to engaging the delay discounting target, we aimed to describe how EFT may be functioning using a mathematical instantiation of the competing neurobehavioral decision systems (CNDS) theory. The CNDS model was developed to understand the challenges in self-regulation and self-control. This conceptual model posits that in substance use disorder, choice results from an imbalance between two decision systems (impulsive and executive) which are functionally linked to limbic and executive neural circuits (Bickel et al., 2011a, 2007; McClure and Bickel, 2014). In particular, individuals with cocaine use disorder demonstrate hyperactivity in impulsive neural circuits, and hypoactivity in executive neural circuits (Bickel et al., 2016a). When performing a discounting task in an MRI scanner, these circuits are active when choosing a short or long term outcome, respectively (McClure et al., 2004).

The model was initially developed to summarize the interaction of these neural systems (McClure et al., 2007). The mathematical form of the model captures the idea that impulsive (β) and executive (δ) systems differ in their effective discount rate (β<δ) and that decision making depends on input from both systems (through a weighted sum captured by ω). The β-δ model is the most simplistic mathematical equation that captures these hypotheses. Additionally, this model approximates the hyperbolic model very closely. We expect that future examination of the mechanisms producing the temporal window will require that the model of the CNDS that we employed will require modification.

Despite the simplicity of the model as an approximation of the CNDS theory, we believe that fitting the model in this experiment yielded important insights into the mechanisms by which EFT influenced behavior. We found that a single EFT session leaves the executive (δ) and impulsive (β) systems unchanged, but that EFT did increase the degree to which the executive system is used in evaluating future rewards (ω). This has potentially profound clinical implications. If the components for evaluation of near- (β) and far-term (δ) consequences are impervious to change, then this suggests that clinical interventions should not target how people conceive of immediate and future consequences per se. Instead, the relative reliance on evaluation systems (ω) is more amenable to change, according to our analysis. Thus, while acknowledging immediate and future rewarding outcomes, participants may be encouraged to attend more to future outcomes when making decisions. This would increase ω and decrease discount rates (k) to reduce impulsivity, regardless of whether the individual demonstrates hyperactivity of the near-sighted system, hypoactivity of the far-sighted system, or both.

In terms of neural targets, it is unclear what brain areas may determine ω. One possibility that we favor is in ventromedial prefrontal cortex (vmPFC) which receives input from executive control regions in dorsolateral PFC as well as input from limbic regions important for reward processing. Because vmPFC sits at this anatomical interface, it has been proposed to be critical for integrating affective and cognitive information for decision-making (Eslinger and Damasio, 1985). Dysfunction of vmPFC is thought to underlie problems in emotion regulation, such as depression, for similar anatomically-based reasons (Mayberg et al., 1997).

An important limitation must be noted in our analysis of discounting data using the CNDS model. Figure 3 illustrates that ω changed significantly with the EFT manipulation and that it did not change significantly following the control ERT manipulation. However, the change in ω is not significantly different when comparing the effects of the EFT and ERT manipulations. Interpretation of the modeling results are therefore only relevant secondary to our primary analysis showing that discounting was differentially affected in EFT and ERT using the more sensitive ANOVA tests of AUClogd measures of discounting. The modeling results provide additional insight into why discounting changed. The power of the modeling analyses is reduced relative to the use of ANOVA since the model requires estimating multiple parameters that interact non-linearly. Future studies that specifically aim to use the CNDS model to characterize discounting should be mindful of the reduced power of this analytic approach. For the current study, it is important to emphasize that we use the CNDS model for secondary analysis of a statistically significant effect to gain insight into possible mechanisms.

From the Experimental Medicine perspective, once engagement of a clinically-relevant target is possible, testing for long-term behavioral change should be sought. Our data indicate that targeting an individual’s temporal window, the time over which rewards are integrated, not only changes delay discounting but also influences demand for drugs. We believe that this effect may be made to persist long-term. For example, discount rates in abstaining smokers remained low 12 months after a 6-week cognitive behavioral therapy intervention (Secades-Villa et al., 2014). Neurally, developmental changes in delay discounting are associated with concomitant changes in executive and limbic brain systems (Van Den Bos et al., 2015). Brain plasticity persists throughout life, so that these changes may be subject to interventions implemented during adulthood.

The present studies demonstrate an interesting and impactful change of perspective and behavior when cocaine users are exposed to narrative interventions. As with most human studies, some considerations must be acknowledged while interpreting the outcomes. First, a total of 5 participants were excluded from analyses due to a violation of Criterion 2 in the money-money task. Given that this is a reasonably large proportion of the collected sample, we acknowledge that the possibility exists that some participants may not have adequately understood or attended to the tasks. In contrast, individuals who were paying close attention to the task may have demonstrated demand characteristics (Rung and Madden, 2018). However, recent work has shown that even if participants can identify the experimental group they are in, the effect of EFT on both delay discounting and demand remains significant even when controlling for demand characteristics (Rung and Madden, 2019; Stein et al., 2018).

5. Conclusions

In conclusion, episodic future thinking is a potential intervention that engages the target of reinforcer pathology. From the perspective of the Experimental Medicine Approach, future potential therapeutic techniques which actively engage reinforcer pathology as a target may also produce clinically relevant outcomes. Future work should focus not only identifying new potential interventions to engage reinforcer pathology as a target, and use the clinically relevant outcomes as a litmus test for mechanism.

Supplementary Material

1

Highlights.

  • The temporal window is a potential underlying mechanism for substance use disorder

  • Episodic future thinking decreases discount rates and demand for cocaine

  • The Experimental Medicine Approach may promote substance use intervention development

Acknowledgements

This work was supported by R01 AA027381 and the Fralin Biomedical Research Institute. The results from both studies were previously published in two distinct abstracts. First, Study 1 results were presented as an oral presentation at the 2018 annual College on Problems of Drug Dependence conference in [Drug & Alcohol Dependence. 2018 “Stimulating future value: Episodic future thinking decreases delay discounting in recreational and chronic cocaine users” by S.E. Snider, J.K.Turner, and W.K. Bickel]. Second, Study 2 results were presented in poster format at the 2019 annual meeting of the College on Problems of Drug Dependence conference in San Antonio, Texas [Drug & Alcohol Dependence. 2019, Abstract No. 73. “Exacerbating reinforcer pathology: Hurricane-associated loss increases delay discounting and demand for cocaine in cocaine users” by S.E. Snider, J.K. Turner, and W.K. Bickel].

Footnotes

Conflict of Interest

Although the following relationships do not create a conflict of interest pertaining to this manuscript the authors report the following. S. E. Snider and W. K. Bickel are principals of BEAM Diagnostics, Inc. W.K. Bickel is also a principal of HealthSim, LLC; Notifius, LLC; and Red 5 Group, LLC. In addition, he serves on the scientific advisory board for Sober Grid, Inc. and is a consultant for Alkermes, Inc.. J. K. Turner and S. M. McClure have no disclosures.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Amlung M, Vedelago L, Acker J, Balodis I, MacKillop J, 2016. Steep delay discounting and addictive behavior: a meta-analysis of continuous associations. Addiction. 10.1111/add.13535 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Atance CM, O’Neill DK, 2001. Episodic future thinking. Trends Cogn. Sci 5, 533–539. [DOI] [PubMed] [Google Scholar]
  3. Beck AT, Epstein N, Brown G, Steer RA, 1988. An inventory for measuring clinical anxiety: psychometric properties. J. Consult. Clin. Psychol 56, 893–897. [DOI] [PubMed] [Google Scholar]
  4. Beck AT, Steer RA, Brown GK, Others, 1996. Manual for the beck depression inventory-II. San Antonio, TX: Psychological Corporation; 1, 82. [Google Scholar]
  5. Bernard C, 1957. An Introduction to the Study of Experimental Medicine, by C.Bernard. Trans. from the French by H.C.Green. [Google Scholar]
  6. Bickel WK, Jarmolowicz DP, Mueller ET, Gatchalian KM, 2011a. The behavioral economics and neuroeconomics of reinforcer pathologies: implications for etiology and treatment of addiction. Curr. Psychiatry Rep 13, 406–415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bickel WK, Johnson MW, Koffarnus MN, MacKillop J, Murphy JG, 2014a. The behavioral economics of substance use disorders: reinforcement pathologies and their repair. Annu. Rev. Clin. Psychol 10, 641–677. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Bickel WK, Koffarnus MN, Moody L, Wilson AG, 2014b. The behavioral- and neuro-economic process of temporal discounting: A candidate behavioral marker of addiction. Neuropharmacology 76 Pt B, 518–527. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bickel WK, Landes RD, Christensen DR, Jackson L, Jones BA, Kurth-Nelson Z, Redish AD, 2011b. Single- and cross-commodity discounting among cocaine addicts: the commodity and its temporal location determine discounting rate. Psychopharmacology 217, 177–187. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Bickel WK, Miller ML, Yi R, Kowal BP, Lindquist DM, Pitcock JA, 2007. Behavioral and neuroeconomics of drug addiction: competing neural systems and temporal discounting processes. Drug Alcohol Depend. 90 Suppl 1, S85–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Bickel WK, Snider SE, Mellis AM, 2019. Using an Experimental Medicine Approach to Identify Novel Determinants of Addiction. Perspectives on Behavior Science 42, 385–396. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Bickel WK, Snider SE, Quisenberry AJ, Stein JS, Hanlon CA, 2016a. Competing neurobehavioral decision systems theory of cocaine addiction: From mechanisms to therapeutic opportunities. Prog. Brain Res 223, 269–293. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Bickel WK, Wilson AG, Chen C, Koffarnus MN, Franck CT, 2016b. Stuck in Time: Negative Income Shock Constricts the Temporal Window of Valuation Spanning the Future and the Past. PLoS One 11, e0163051. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Borges AM, Kuang J, Milhorn H, Yi R, 2016. An alternative approach to calculating Area-Under-the-Curve (AUC) in delay discounting research: ALTERNATIVE TO AUC. J. Exp. Anal. Behav 106, 145–155. [DOI] [PubMed] [Google Scholar]
  15. Bruner NR, Johnson MW, 2014. Demand curves for hypothetical cocaine in cocaine-dependent individuals. Psychopharmacology 231, 889–897. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Daniel TO, Stanton CM, Epstein LH, 2013a. The future is now: comparing the effect of episodic future thinking on impulsivity in lean and obese individuals. Appetite 71, 120–125. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Daniel TO, Stanton CM, Epstein LH, 2013b. The future is now: reducing impulsivity and energy intake using episodic future thinking. Psychol. Sci 24, 2339–2342. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Du W, Green L, Myerson J, 2002. Cross-cultural comparisons of discounting delayed and probabilistic rewards. Psychol. Rec 52, 479. [Google Scholar]
  19. Eslinger PJ, Damasio AR, 1985. Severe disturbance of higher cognition after bilateral frontal lobe ablation: patient EVR. Neurology 35, 1731–1741. [DOI] [PubMed] [Google Scholar]
  20. Johnson MW, Bickel WK, 2008. An algorithm for identifying nonsystematic delay-discounting data. Exp. Clin. Psychopharmacol 16, 264–274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Koffarnus MN, Bickel WK, 2014. A 5-trial adjusting delay discounting task: accurate discount rates in less than one minute. Exp. Clin. Psychopharmacol 22, 222–228. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Koffarnus MN, Franck CT, Stein JS, Bickel WK, 2015. A modified exponential behavioral economic demand model to better describe consumption data. Exp. Clin. Psychopharmacol 23, 504–512. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. MacKillop J, Amlung MT, Few LR, Ray LA, Sweet LH, Munafò MR, 2011. Delayed reward discounting and addictive behavior: a meta-analysis. Psychopharmacology 216, 305–321. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Mayberg HS, Brannan SK, Mahurin RK, Jerabek PA, Brickman JS, Tekell JL, Silva JA, McGinnis S, Glass TG, Martin CC, Fox PT, 1997. Cingulate function in depression: a potential predictor of treatment response. Neuroreport 8, 1057–1061. [DOI] [PubMed] [Google Scholar]
  25. McClure SM, Bickel WK, 2014. A dual-systems perspective on addiction: contributions from neuroimaging and cognitive training. Ann. N. Y. Acad. Sci 1327, 62–78. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. McClure SM, Ericson KM, Laibson DI, Loewenstein G, Cohen JD, 2007. Time discounting for primary rewards. J. Neurosci 27, 5796–5804. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. McClure SM, Laibson DI, Loewenstein G, Cohen JD, 2004. Separate neural systems value immediate and delayed monetary rewards. Science 306, 503–507. [DOI] [PubMed] [Google Scholar]
  28. Mellis AM, Snider SE, Bickel WK, 2018. Narrative theory: II. Self-generated and experimenter-provided negative income shock narratives increase delay discounting. Exp. Clin. Psychopharmacol 26, 113–118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. O’Neill J, Daniel TO, Epstein LH, 2016. Episodic future thinking reduces eating in a food court. Eat. Behav 20, 9–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Riddle M, Science of Behavior Change Working Group, 2015. News from the NIH: using an experimental medicine approach to facilitate translational research. Transl. Behav. Med 5, 486–488. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Rung JM, Madden GJ, 2019. Demand characteristics in episodic future thinking II: The role of cues and cue content in changing delay discounting. Exp. Clin. Psychopharmacol 27, 482–495. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Rung JM, Madden GJ, 2018. Demand characteristics in episodic future thinking: Delay discounting and healthy eating. Exp. Clin. Psychopharmacol 26, 77–84. [DOI] [PubMed] [Google Scholar]
  33. Rung JM, Peck S, Hinnenkamp JE, Preston E, Madden GJ, 2019. Changing Delay Discounting and Impulsive Choice: Implications for Addictions, Prevention, and Human Health. Perspectives on Behavior Science. 10.1007/s40614-019-00200-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Secades-Villa R, Weidberg S, García-Rodríguez O, Fernández-Hermida JR, Yoon JH, 2014. Decreased delay discounting in former cigarette smokers at one year after treatment. Addict. Behav 39, 1087–1093. [DOI] [PubMed] [Google Scholar]
  35. Snider SE, LaConte SM, Bickel WK, 2016a. Episodic Future Thinking: Expansion of the Temporal Window in Individuals with Alcohol Dependence. Alcohol. Clin. Exp. Res 40, 1558–1566. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Snider SE, LaConte SM, Bickel WK, 2016b. REMEMBER TO THINK AHEAD: WORKING MEMORY TRAINING IMPROVES EPISODIC FUTURE THINKING RATE DEPENDENTLY IN ALCOHOL-DEPENDENTS, in: ALCOHOLISM-CLINICAL AND EXPERIMENTAL RESEARCH. WILEY-BLACKWELL 111 RIVER ST, HOBOKEN 07030–5774, NJ USA, p. 125A–125A. [Google Scholar]
  37. Snider SE, Mellis AM, Poe LM, Kocher MA, Turner JK, Bickel WK, 2019. Reinforcer pathology: Narrative of hurricane-associated loss increases delay discounting, demand, and consumption of highly palatable snacks in the obese. Psychol. Addict. Behav 10.1037/adb0000498 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Stein JS, George Wilson A, Koffarnus MN, Daniel TO, Epstein LH, Bickel WK, 2016. Unstuck in time: episodic future thinking reduces delay discounting and cigarette smoking. Psychopharmacology. 10.1007/s00213-016-4410-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Stein JS, Koffarnus MN, Snider SE, Quisenberry AJ, Bickel WK, 2015. Identification and management of nonsystematic purchase task data: Toward best practice. Exp. Clin. Psychopharmacol 23, 377–386. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Stein JS, Sze YY, Athamneh L, Koffarnus MN, Epstein LH, Bickel WK, 2017. Think fast: rapid assessment of the effects of episodic future thinking on delay discounting in overweight/obese participants. J. Behav. Med 10.1007/s10865-017-9857-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Stein JS, Tegge AN, Turner JK, Bickel WK, 2018. Episodic future thinking reduces delay discounting and cigarette demand: an investigation of the good-subject effect. J. Behav. Med 41, 269–276. [DOI] [PubMed] [Google Scholar]
  42. Tiffany ST, Drobes DJ, 1991. The development and initial validation of a questionnaire on smoking urges. Br. J. Addict 86, 1467–1476. [DOI] [PubMed] [Google Scholar]
  43. Van Den Bos W, Rodriguez CA, Schweitzer JB, McClure SM, 2015. Adolescent impatience decreases with increased frontostriatal connectivity. Proceedings of the National Academy of Sciences 112, E3765–E3774. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. van den Bos W, Rodriguez CA, Schweitzer JB, McClure SM, 2014. Connectivity strength of dissociable striatal tracts predict individual differences in temporal discounting. J. Neurosci 34, 10298–10310. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Watson D, Clark LA, Tellegen A, 1988. Development and validation of brief measures of positive and negative affect: the PANAS scales. J. Pers. Soc. Psychol 54, 1063–1070. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

1
NIHMS1711148-supplement-1.docx (31.5KB, docx)

RESOURCES