Delay discounting refers to the tendency for the value of a commodity (e.g., money) to decrease as the delay to its receipt increases. The rate at which a commodity loses its value (i.e., discounting rate) is determined by asking participants to choose repeatedly between “smaller-sooner” and “larger-later” outcomes. In general, smaller-sooner choices are viewed as “impulsive,” whereas larger-later choices are viewed as “self-controlled.” Because exaggerated preference for smaller-sooner outcomes (i.e., higher discounting rates) is commonly associated with impulsive behaviors such as substance abuse, pathological gambling, and sexual risk behavior (Amlung et al., 2017), researchers have sought to identify methods for decreasing discounting rates (i.e., promoting greater larger-later choices, or self-control).
Notably, episodic future thinking (EFT) has been shown to decrease discounting rates (Peters & Büchel, 2010; Stein et al., 2017). One explanation as to why EFT is effective is that by imagining future events in vivid detail, individuals may be more likely to value delayed outcomes in the present when making choices that will benefit their future selves. In a similar way, individuals may construct vivid imagery by recalling past events. The possibility that episodic past thinking (EPT) could achieve the same positive benefit as EFT prompted Daniel and colleagues in 2016 to compare the effects of EFT vs. EPT on delay discounting. In their study, participants who were randomly assigned to engage in EFT or EPT ( n = 27 per group) discounted hypothetical monetary rewards significantly less steeply than participants assigned to a control condition (episodic recent thinking, or ERT). As potentially important as these findings are for the development of episodic thinking treatments, they have yet to be reproduced. Given recent concerns within psychology about failures to reproduce high-profile empirical findings, the present study had two goals. In Experiment 1, we attempted to replicate the procedures of Daniel et al. (2016) as closely as possible with college students. In Experiment 2, we sought to examine the potentially beneficial effects of episodic thinking on delay discounting in a population vulnerable to impulsive choice, namely individuals with DSM-5 Alcohol Use Disorder (AUD; Petry, 2001). Both experiments were reviewed and approved by the Chico State Institutional Review Board (IRB).
Experiment 1 successfully adapted the procedures used by Daniel et al. (2016) into an online survey format with college students. Eighty-one participants were randomly assigned to EFT ( n = 24), EPT ( n = 29), or ERT ( n = 28) conditions. Once assigned to their episodic cue condition, participants were prompted to generate personally relevant episodic cues using instructional sets provided by the authors of Daniel et al. (2016). Next, participants completed two delay discounting tasks: one involving past rewards and one involving future rewards. During the discounting tasks, episodic cues generated earlier in the survey were displayed above the choice alternatives. Although we reproduced Daniel et al.’s (2016) findings that EFT and EPT episodic cues were significantly more salient and arousing relative to ERT episodic cues, we did not reproduce the positive effects of EFT and EPT on delay discounting relative to the ERT control condition.
Experiment 2, which began recruitment during Experiment 1, sought to enroll AUD and non-AUD (control) participants through Amazon Mechanical Turk, a crowdsourcing platform commonly used by academic researchers to conduct surveys. Inclusion criteria for the AUD group was consistent with criteria from similar research: Individuals (≥18 years old) needed to (1) endorse ≥ 4 DSM-5 criteria for AUD (moderate to severe AUD); and (2) endorse < 2 DSM-5 criteria for other substances (except nicotine and tobacco). To qualify for the non-AUD group, individuals needed to endorse < 2 DSM-5 criteria for all substances (except nicotine and tobacco). Unfortunately, of the 107 individuals who completed the screening questionnaire, only 5 met inclusion criteria for the AUD group and 7 met inclusion criteria for the non-AUD group. Based on the rate of recruitment, we estimated that it would take considerably longer than expected to recruit the desired sample ( N = 180; n = 90 per group). The experiment was therefore discontinued.