Contribution
The first test of an acceptance-based intervention with delay discounting as the target. One session of ACT-derived exercises and values work lowered measured discounting a week later in students selected for steep discounting, relative to a waitlist. The process measures the training was meant to move, psychological flexibility and distress tolerance, did not change relative to control.
Key insight
If choosing the sooner reward is partly escape from the discomfort of waiting, then teaching people to have that discomfort without acting on it should make them wait more. Here the choices moved toward waiting. The questionnaire measure of the willingness that was supposed to explain the move did not.
Assumptions
- Choosing the SSR as avoidance. “A mechanism of delay discounting is that selection of the SSR represents avoidance of the aversive properties of the delay to the LLR” (Introduction). The abstaining smoker’s withdrawal and craving are the example.
- Distress intolerance and psychological inflexibility are related constructs that acceptance-based treatment targets (Introduction).
- Hypothetical money stands for real choice. The discounting task (after Rachlin et al. 1991) uses a hypothetical $1000 at seven delays from 1 week to 25 years, with test–retest r = .91 at one week as cited, and r = .953 in this sample (Measures; Discussion).
- The AAQ-II measures the process. Seven items, “a 7-item Likert-scale measure of experiential avoidance and psychological inflexibility”; scores above 24–28 go with clinical distress (Measures).
Key results
- Sample (Participants). Utah State undergraduates screened online. Those with AUC < 0.4 and systematic data were invited. 34 were randomized, 17 per group. Four withdrew, one because “the training elicited distress”. 30 completed (ages 18–41, mean 21.5, half women, 90% without meditation experience). Two per analysis were excluded for nonsystematic discounting data.
- Training (Acceptance-Based Training Session). One 60–90 minute session with a graduate student under a licensed psychologist’s supervision, on “acceptance and values”. It included a leaky-pipe exercise on the futility of control, the Polygraph and Chocolate Cake exercises, the Unwanted Party Guest metaphor, and carrying a card with an unwanted thought on it all day. Participants who reacted with fear were helped to view “the thought as a thought (i.e., private behavior without literal meaning)” by holding an object in an open hand, which “illustrate[s] the distinction between the person holding the thought and the thought itself”. Then came a values bull’s-eye, a fork-in-the-road variant of Passengers on the Bus, and a week of goal-tracking. The training opened by telling participants that “impulsive decision making may cause problems for you”.
- Time 1 to Time 2, trained against waitlist (Results, Table 1). AUC went from 0.26 (SD 0.16) to 0.44 (0.22) in the trained group and from 0.22 (0.13) to 0.25 (0.17) on the waitlist. Group × time F(1, 26) = 4.658, p = .04, partial η² = .152. Time F(1, 26) = 10.725, p = .003. Group F(1, 26) = 4.098, p = .053.
- Process measures did not move (Results). AAQ-II: time p = .202, training p = .281, interaction p = .377. The means were 23.27 to 21.47 trained and 19.07 to 18.73 waitlist (lower is more flexible). DTS: time p = .092, training p = .447, interaction p = .491.
- The waitlist after its own training (Results). DTS rose, t(14) = −2.49, p = .026. AUC did not change (p = .267) and the AAQ-II did not either (p = .132).
- Pooled pre–post (Results, Table 2). AUC 0.27 to 0.39, t(25) = −2.911, p = .007, d = .57. DTS 3.46 to 3.66, t(29) = −2.665, p = .012, d = .49. AAQ-II 21.00 to 19.50, t(29) = 1.972, p = .058, d = .36.
- The authors’ reading (Discussion). “One potential interpretation is that distress tolerance plays a larger role than psychological flexibility in delay discounting.” They grant that the pooled analysis has no control group.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | A brief acceptance-based session reduces delay discounting of hypothetical money a week later, relative to a waitlist | weak to moderate | interaction p = .04 in 28 analysed; not replicated within the waitlist group’s own training (p = .267) |
| C2 | The session does not change psychological inflexibility (AAQ-II) or distress tolerance relative to a waitlist | weak, as a null | n = 30; no interaction for either |
| C3 | Pooled across groups, training improves discounting and distress tolerance | weak | pre–post with no control |
| C4 | Distress tolerance may matter more than psychological flexibility for discounting | speculation | the authors’ “one potential interpretation” |
Concepts
- AUC — the area under the curve of indifference points, from 0 (always the sooner reward) to 1 (no discounting).
- Psychological inflexibility, experiential avoidance — what the AAQ-II measures, after Bond et al. 2011.
- Distress tolerance — the DTS’s beliefs about tolerating, appraising, being absorbed by and regulating distress (Simons & Gaher 2005).
Connections
It applies the experiential-avoidance account (Hayes et al. 1996) to Ainslie’s impulsive choice. Madden and Odum’s laboratory supplies the discounting methods (AUC; nonsystematic-data criteria of Johnson and Bickel 2008). Rung and Madden’s meta-analysis (LIT-1846, NOTE-1440) includes it as one of two mindfulness-based studies, with g = 0.94. That review describes the between-group difference as trend-level (p < .06). The paper reports the group × time interaction at p = .04, and it is the group main effect that is p = .053. The review’s figure is an effect size it computed, and the difference is in which test is meant.
Bearing on the record
- CLAIM-156: against it in part, without meeting its defeat condition. The claim says mindfulness integrates by suspending the step from impulse to action, a step “that needs no change in the discount curves”. Here an acceptance session, with a “thought as a thought” exercise for some participants, lowered measured discounting. So some of its effect was on the valuations, which is the flow route the claim’s body says Dixon et al. (LIT-906) would point to. The study does not meet the defeat condition. It measures only discounting, not self-control or choice under temptation apart from it, so it cannot show the effect fully mediated by discounting. It is also mostly acceptance and values, not mindfulness as the claim describes it. And it is one pilot, whose waitlist arm did not replicate the change after its own training.
- THEORY-050: against its process claim, weakly, and outside its
promote_when. The outcome changed and the AAQ-II, ACT’s measure of inflexibility, did not change relative to control. It only approached significance when pooled without a control. So nothing here shows inflexibility changing first or carrying the effect. The authors themselves suggest distress tolerance may matter more. But the AAQ-II is the global self-reportpromote_whensets aside, the sample is non-clinical (mean AAQ-II 19 to 23, near or below the clinical range), and n = 30. This is weak evidence against inflexibility as the process here, not a test of the account. - THEORY-051. The training’s object-in-hand exercise separates “the person holding the thought” from “the thought itself”, a self-as-context move. It was given only to some participants and not measured separately, so it bears on nothing the account waits on.
- No instruction for machine-learning practice; nothing for the anthology.
Limitations
The authors list six (Discussion):
- A Hawthorne effect, since participants were told the study’s rationale before training but not during the waitlist.
- A small pilot sample.
- A homogeneous sample: mostly White undergraduates.
- Mostly subclinical problems.
- No adherence coding of the session.
- No long-term follow-up.
Three more:
- The waitlist group’s post-training discounting change was not significant (p = .267). The pooled pre–post effect leans on the trained group.
- Discounting was measured with hypothetical money, and the one additional discounting measure that was given is “not reported in this paper” (Measures).
- The intervention was tailored to each participant’s “particular struggles”, so the manipulation is not uniform.
Open questions
- Against an active or sham control, and with real or non-monetary rewards, does the effect on discounting survive?
- Does anything measured, whether distress tolerance, the AAQ-II or a behavioural measure of avoidance, change before discounting and mediate it?
- Does the training change self-control in choices under temptation by more than its change in discounting carries? That is the test CLAIM-156 asks for.