It is a truism that most smokers want to quit and that most who make a quit attempt relapse even if they receive the best available psychosocial and/or pharmacological treatments (Fiore et al. 2008
). As such, there is considerable scope for improving the efficacy of interventions for quitting smokers. This might involve relatively minor modifications to existing treatment approaches or a more radical shift in the way smokers are routinely helped to quit (Kamboj and Das 2017
Existing behavioural interventions often encourage smokers to reduce exposure to relapsogenic situations or cues and/or promote the regulation of affect (including craving) through the use of strategies intended to modify the content, frequency or intensity of private psychological experiences. For example, smokers learn to identify and systematically dispute unhelpful thoughts (e.g. ‘one cigarette won’t hurt’). These strategies belong to a family of behavioural therapies—variously termed behavioural counselling, coping skills training or cognitive behavioural therapy (CBT)—that generally involve antecedent-focused emotion regulation strategies (which modulate emotions at input) like reappraisal or situation modification (e.g. avoiding places where smoking is likely). A primary assumption of these approaches is that regulation of (reduction in) craving and/or negative affect is required before behaviour change occurs.
However, some researchers have pointed to limitations of such cognitive-behavioural emotion regulation strategies. For example, the ubiquity of smoking cues, their frequent occurrence in both the external and the less easily regulated internal (bodily) environments, makes it impossible to avoid them. Moreover, attempts to apply certain response-focused emotion regulation strategies (which are deployed once an emotional response has commenced, e.g. suppression), to dampen craving, negative affect or smoking-related thoughts, can result in rebound effects (Salkovskis and Reynolds 1994
; Sayers and Sayette 2013
; Toll et al. 2001
). More generally, strategies that emphasise control over internal experiences (thoughts and feelings) are likely only to be successful in those with significant pre-existing cognitive reserves, although even these individuals are susceptible to the ironic effects of over-control, especially under conditions of stress (Garland et al. 2014
; Muraven and Baumeister 2000
In contrast, advocates of recently developed smoking cessation interventions informed by contextual behavioural science consider efforts to downregulate private subjective experiences to be unrealistic, offering only temporary relief from unwanted experiences and generating an unworkable problem-solving mindset towards such experiences (Blackledge and Hayes 2001
). They suggest an alternative approach to understanding addiction that is gaining influence, namely the psychological flexibility model (PFM), the principles of which are embodied and applied in Acceptance and Commitment Therapy (ACT; Hayes et al. 2011
). The most relevant psychological constructs in ACT and other PFM-consistent interventions are self-regulated attention and openness or orientation to experience, characterised by acceptance of, and curiosity towards, current experience (Bishop et al. 2004
). Interested readers are referred to authoritative texts on the PFM, which describe its underlying theory and therapeutic components in detail (e.g. Hayes et al. 1999
Unlike CBT, which assumes that thoughts have some inherent truth value that can be disputed, the PFM proposes that cognitive disputation strategies give too much weight to thoughts and can promote maladaptive responding (cognitive fusion). Alternatively, therapeutic strategies that encourage an experiential understanding of the essentially illusory and transient nature of thoughts and feelings are foundational techniques in ACT and mindfulness-based addiction therapies (Hayes et al. 2011
). In essence, changing the way individuals relate to their internal experiences (i.e. observing them with curiosity and accepting them as temporary mental processes that are not inherent to the self) can have the effect of discouraging experiential avoidance strategies, which are considered primary transdiagnostic maintaining factors in psychological/substance use disorders.
Whilst regulation of the intensity or occurrence of private psychological experiences is not the focus of PFM interventions, reductions in craving and negative affect may nonetheless occur indirectly and secondarily to increased attention towards (or reduced avoidance of) such experiences (Farb et al. 2014
; Witkiewitz et al. 2013
). This could occur, for example, through enhanced extinction of conditioned craving responses, during which simultaneous exposure to, and attention towards, multiple conditioned drug stimuli (in the absence of drug reward) could drive greater inhibitory learning (e.g. Treanor 2011
). In contrast, ‘ordinary’ extended cue exposure has limited effects on smoking-related cue reactivity (subjective craving and physiological responding), even when combined with drugs that ostensibly enhance extinction learning (e.g. Kamboj et al. 2012
). Alternatively, mindfulness-like strategies could interrupt (compete with) the elaboration of desire-related thoughts, with knock-on effects on craving (Farb et al. 2014
; May et al. 2011
). To reiterate, however, direct attempts to moderate the intensity of internal experiences are not employed in PFM-guided interventions and would not feature in the rationale for their use as presented to treatment recipients. Indeed, given the paradoxical nature of instructions to approach and ‘stay with’ sensations of craving or negative affect, provision of a clear rationale and, preferably, experiential exercises to exemplify the purpose of these strategies would seem to be essential to ensuring compliance with techniques that may, at first blush, be experienced as counterintuitive (Levin et al. 2012
ACT is a multi-component treatment. Four of the components of ACT—contact with the present moment, acceptance, defusion and self-as-context—fall under the umbrella of mindfulness and acceptance practices (values and committed action are its fifth and sixth elements; Hayes et al. 2011
). In contrast, mindfulness-based relapse prevention (MBRP) is monomodal insofar as the central therapeutic activities are various mindfulness meditation exercises (Bowen et al. 2014
). Although mindfulness is conceptualised somewhat differently in ACT than in MBRP and related mindfulness treatments, we argue that there is sufficient convergence between these approaches (in terms of inculcating a particular attitude to the here and now and in aspects of the definition of mindfulness) to consider their constituents to belong to the same family of PFM-consistent strategies (or, simply, PFM strategies).
Evidence for efficacy of comprehensive PFM interventions for smoking cessation is accumulating. However, clinical trials are not ideally suited to testing the theoretical bases of the PFM. Rather, experimental laboratory studies that examine isolated components of these interventions can be more useful in this regard. Moreover, such lab-based studies allow the safety and relative efficacy of treatment components to be efficiently and cost-effectively tested. In this way, the most promising components can be retained and combined into optimised treatment packages and/or tested in more complex factorial designs (Hayes et al. 2013
). Recently, Levin et al. (2012
) reviewed laboratory studies that examined the effects of isolated components of the PFM, providing support for the efficacy of individual components across a broad range of participants and problem types. However, at the time of that review, few relevant studies had been conducted in smokers. Although previous studies have reviewed mind-body- and mindfulness-based treatments for smoking cessation either systematically (Carim-Todd et al. 2013
; de Souza et al. 2015
) or narratively (McCallion and Zvolensky 2015
), no review that we are aware of has specifically examined the variety of components of the model and focused specifically on experimental component studies.
We systematically review laboratory studies of PFM strategies relevant to smoking cessation. Whilst we examine outcomes with established relevance to smoking cessation—specifically craving and negative affect—the more relevant outcomes from the perspective of the PFM include smokers’ relationship to these experiences (Levin et al. 2012
), cessation-relevant values and, importantly, cessation-relevant behaviours. Whilst other psychosocial therapies would obviously also view a reduction in smoking as the primary goal of treatment, as noted above, behaviour change in these approaches is expected to be mediated by changes in negative affect or craving and/or smoking-related beliefs. In contrast, the PFM predicts a reduction in smoking independently of reductions in craving or negative affect (Brewer et al. 2013
; Elwafi et al. 2013
). Given these distinct goals of PFM strategies on one hand, and emotion regulation-based cognitive/behavioural strategies on the other, it should be clear that distinct outcome measures that tap these differing goals should ideally be used in experimental studies that assess these respective approaches. Specifically, whereas craving and negative affect are likely to be important outcomes in assessing emotion regulation-based approaches, measures of meta-cognitive/affective processes, values and/or behaviour change are more relevant to studies of PFM strategies (Levin et al. 2012
). It was therefore of interest to determine the extent to which the most suitable outcome measures were routinely used in studies of PFM strategies in smokers. As craving, negative affect and smoking behaviour were the most commonly assessed constructs in the identified studies, provisional aggregate effect sizes are reported for these when data was (made) available.
In this review, we provide a systematic synthesis of findings from, and methodological characteristics of, studies examining the isolated effects of PFM strategies that are used as therapeutic ingredients in PFM interventions for smoking cessation. Where sufficient data were (made) available, ESs were reported, although ESs for PFM outcomes were based on a small number of studies and therefore lacked the precision to make a robust determination of strategy effects. As it stands, the qualitative and limited quantitative findings did not indicate beneficial effects of specific, brief and isolated PFM strategies for smoking behaviour. In contrast, PFM strategy use was associated with reductions in acute craving and negative affect when compared to inactive controls, but not active strategies, which is the more stringent comparison.
Before proceeding, it is worth reemphasising that the purpose of controlled experimental laboratory component studies is rarely to establish clinical efficacy, but rather to develop insights about the validity of theoretical and psychotherapeutic concepts that can then be used to refine theory or streamline treatment (e.g. by identifying potentially inert or weakly effective components; Hayes et al. 2011
). To be informative in this regard, clear theoretical foundations and high levels of internal validity should be brought to bear. However, all of the studies reviewed here have some limitations in this respect, which currently leave some uncertainty about their true impact on emotion regulation and PFM-specific outcomes.
Amongst the most significant limitations of the reviewed studies was a lack of theoretically driven predictions relating to the effects of PFM strategies (i.e. predictions that were clearly grounded in the PFM/mindfulness theory) and limited use of appropriate (PFM-consistent) assessment measures. Some of the reviewed studies seemed, either implicitly or explicitly, to consider the PFM strategies to be emotion regulation
techniques, that is, strategies geared towards controlling (reducing) the occurrence or intensity of unwanted or maladaptive internal experiences. For example, one study stated that according to ACT theory, a larger reduction in negative affect and craving following acceptance instructions was expected relative to comparison conditions including reappraisal (Szasz et al. 2012
). However, this prediction is not, in fact, consistent with PFM theory. Another study, whilst providing explicit differential hypotheses for reappraisal and defusion, based on emotion regulation and ACT theories, respectively, nonetheless described both as emotion regulation strategies (Beadman et al. 2015
). These conceptual issues can have consequences for decisions on study design.
For example, the tendency to conceptualise PFM techniques as emotion regulation strategies may partly explain why only six studies examined outcomes more suited to testing PFM strategy effects, but all examined negative affect and/or craving. Several of these provided specific hypotheses (although the theoretical bases for these hypotheses were generally unclear) in relation to the latter outcomes, generally predicting and reporting acute reductions in intensity of craving in the PFM strategy groups. Yet, unlike emotion regulation-based (cognitive-behavioural) strategies, which assume that modification of maladaptive thoughts precedes a relatively immediate reduction in the intensity and frequency of feeling states, abrupt reductions in craving after very brief PFM strategy use were less expected. Indeed, it is equally plausible that enhanced interoceptive attention via mindfulness or acceptance instructions would temporarily exacerbate craving and negative mood, particularly when limited explanation of, or experiential practice with, the relevant techniques is provided. However, given that the reviewed studies did tend to show reductions in intensity of subjective experiences, we consider a number of potential explanations for apparent acute emotion regulatory effects of PFM strategies.
Firstly, recent efforts to integrate aspects of the PFM (mindfulness in particular) into the process model of emotion regulation (Farb et al. 2014
) and conceptualise these as unique forms of early-stage attentional regulation may provide testable, mechanistic proposals for “how attention deployment leads to attenuation
of negative emotions” (Farb et al. 2014
, p. 549). For example, mindfulness may feed into reappraisal processes, and enable novel appraisals
of emotional experiences. Related to this, state mindfulness has been shown to mediate mindfulness training effects on reappraisal (Garland et al. 2015b
). Secondly, extended simultaneous processing of multiple cues in the context of a widened scope of attention and approach orientation, in the absence of reinforcement, may reduce conditioned responding through extinction (Treanor 2011
). One line of empirical support for this idea is the finding that reduced drug craving is associated with higher levels of home practice of mindfulness (potentially reflecting greater/more prolonged exposure to conditioned stimuli in a variety of contexts) following a course of treatment with MBRP (Grow et al. 2015
). Thirdly, PFM strategies might disrupt the elaboration of smoking-related cognition (especially mental imagery)—which is implicated in craving—by competing for the limited cognitive resources that normally subserve such elaboration (May et al. 2015
). Finally, reductions in craving and negative affect may simply reflect expectancy effects. Although perhaps the least scientifically appealing, it is difficult to discount this possibility since the mediating effect of expectancy was not generally tested in the reviewed studies, and the between-group differences were most clearly observed under conditions that controlled the least well for it (i.e. when an inactive control was used).
Objective assessments of cue reactivity/craving could help resolve concerns regarding expectancy effects. In this respect, further studies (e.g. Westbrook et al. 2013
) examining the neural circuitry subserving eliminative (e.g. reduction in craving) and potentially generative (e.g. increased positive affect; Garland et al. 2015a
) pathways of PFM strategies might be useful. Moreover, such studies would allow an exploration of how neuroplasticity mediates the effects of PFM strategies on behavioural change (McConnell and Froeliger 2015
) and help us determine the degree to which the neural substrates of PFM strategy effects overlap with those of emotional regulation strategies, like reappraisal. More generally, future studies that focus on the mechanisms through which PFM strategies exert their effects (whether through changes in cognition or affect, or not) would be valuable in helping to test and/or refine theoretical predictions of the PFM and the interventions guided by this model.
In addition, participant expectancy effects could be minimised in future studies by blinding participants to study aims and ensuring that control conditions have similar levels of credibility and expectancy as the PFM condition. Indeed, Beadman et al. (2015
) found that despite attempting to match instructions closely for experimental (defusion), active-control (reappraisal) and suppression-control conditions, participants rated the suppression condition as significantly less credible. As such, an untested assumption of equivalent credibility across experimental and control groups cannot be considered safe. Moreover, no study has yet attempted to limit experimenter
expectancy through blinding. This could be relatively easily achieved through automated experimental procedures and would significantly allay concerns about experimenter allegiance.
Related to issues of credibility and expectancy, the fact that PFM-guided interventions do not primarily aim to reduce symptom intensity may be experienced as counterintuitive by participants (Eifert and Forsyth 2005
). As such, an assessment of participants’ understanding of strategy instructions seems crucial, yet only two of the reviewed studies assessed understanding directly (Litvin et al. 2012
; Nosen and Woody 2013
). Of these, Litvin et al. (2012
) reported poorer understanding of the PFM strategy compared to control instructions. ACT commonly uses metaphors, stories and experiential exercises to illustrate unfamiliar and abstract ideas. Use of such exercises prior to
participants employing the PFM strategy would be an important refinement for future studies, although assessment of comprehension and compliance would still be essential.
In relation to the relatively infrequent use of smoking behaviour as an outcome in the reviewed studies, one potential concern might have been that a measurable reduction in smoking after very brief, single-session instructions was deemed unlikely. Indeed, significant reductions in the number of cigarettes smoked over an extended (several days or weeks) follow-up seem a particularly stringent test of efficacy of PFM strategies. Yet, two of the reviewed studies did indeed show reductions in smoking in the PFM strategy group (Bowen and Marlatt 2009
; Ruscio et al. 2016
). Of these, the reductions in number of cigarettes per day observed by Bowen and Marlatt (2009
) at 7-day follow-up are particularly striking. Their within-session strategy instructions were only 11 min in duration, and participants did not appear to be provided with explicit instructions to practice the PFM strategy (urge surfing) during the follow-up period. This suggests that even extremely brief strategy instructions can have a measurable effect on (self-reported) smoking. However, a number of studies did not include a follow-up assessment period during which smoking was monitored. In these circumstances, alternative approaches to examining smoking behaviour could still be used, and employed within the same approximate time frame as the craving and negative affect assessments. For example, latency to the first cigarette after the experimental session could be assessed (remotely) or a laboratory analogue of relapse could be used (Froeliger et al. 2017
). In addition, smoking topography (i.e. the frequency and duration of puffs) is also a reliable and valid index of changes in smoking behaviour (Lee et al. 2003
). Researchers might therefore consider using these methods to assess smoking behaviour in future single-session experiments testing PFM strategies.
It should be noted that whilst our search terms included all aspects of the PFM, only studies on mindfulness, acceptance and defusion were identified. No appropriate studies of values or committed action based on PFM theory met the criteria, and as such, the effects of these individual intervention components remain unclear. It could be argued that self-affirmation studies should have been included in this review, since these often require participants to recall and reflect upon their values, which can have beneficial effects on drug/alcohol use (e.g. Kamboj et al. 2016
). However, such studies—which typically assess the effects of affirmation on threatened ‘self-integrity’—are derived from social psychological theory developed independently of, and without reference to, the PFM or ACT. Future PFM-informed laboratory experiments focusing on values and committed action will be facilitated by recently developed assessment instruments that are firmly rooted in the PFM.
We must acknowledge some limitations of the current review. The focus of the review was primarily on methodological features of studies. Meta-analysis relies on open science practices, which are becoming increasingly prevalent. A greater availability of data would have strengthened this review by allowing a more comprehensive quantitative reporting to supplement the qualitative appraisal of reviewed studies. On the other hand, the number of available studies would still have been relatively small. As it stands, our findings relating to aggregate ESs were incomplete and potentially susceptible to bias.