MCT builds upon the self-regulatory executive function (S-REF) model, which seeks to explain cognitive and metacognitive factors involved in top-down control and maintenance of psychological disorders (Wells and Matthews, 1996). According to this model, cognitive processes are spread across three interconnected levels: low-level automatic and reflexive processing, cognitive style in the form of conscious processing of thoughts and behaviors, and metacognitive knowledge or beliefs stored in long-term memory. Metacognition refers to awareness and cognition about cognitive processes and includes cognitive factors that control, monitor, and appraise thinking (Wells, 2009).

According to the S-REF model, psychological disorder is linked to a perseverative style of thinking called the cognitive attentional syndrome (CAS; Wells, 2009). CAS consists of prolonged worry or rumination, threat monitoring, and different unhelpful coping styles accompanied by a heightened self-focused attention. This may lead to sustained dysfunctional processing, reduced attentional flexibility, and an experience of uncontrollability of negative thoughts and emotions. The aim in MCT is to eliminate the CAS and to modify dysfunctional metacognitive beliefs about control, appraisal, and cognitive and emotional processing, and thereby strengthen the ability to react in a more flexible way to negative internal stimuli. ATT is one method for achieving this.

A meta-analysis of MCT for anxiety and depression including 16 studies demonstrated large effect sizes and suggested that MCT might be superior to CBT (Normann et al., 2014). Later randomized controlled trials (RCTs) also support the effectiveness of MCT in treating depression (e.g., Jordan et al., 2014; Hagen et al., 2017) and anxiety disorders (e.g., Johnson et al., 2017), and preliminary results indicate that MCT may be suited for group administration (Dammen et al., 2015; Papageorgiou and Wells, 2015).

ATT is a component of MCT designed to strengthen attentional control and promote external focus of attention, to interrupt and break free of the CAS (Wells, 2009). The exercise is auditory, and consists of three sections targeting different attentional components: selective attention, attention switching, and divided attention. The aim is not to distract oneself from difficult thoughts or feelings, but rather to increase flexibility and thus voluntarily being able to choose attentional focus.

Although originally a part of MCT, a growing number of studies are examining the potential of ATT as a standalone intervention. A systematic review with meta-analytic elements summarizes findings from 10 ATT-studies including four studies with a single case experimental design, four RCTs, and two case studies (Knowles et al., 2016). Number of ATT-sessions ranged from 1 to 11 sessions among the included studies. Although still preliminary, the meta-analysis indicates that ATT may be effective in treating a wide range of psychological problems. Four of the included RCTs (Sharpe et al., 2010; Fergus et al., 2014; Nassif and Wells, 2014; Callinan et al., 2015) involved non-clinical samples and used one or two sessions, and are as such comparable to the current study. Within-group effect sizes pre- to post-intervention in these RCTs were medium to large: negative affect (d = 1.03), anxiety measures (range: d = 0.32–0.65), intrusive thoughts (range: d = 1.06–1.33), hypervigilance to pain (d = 0.95), self-focused attention (range: d = 0.55–1.78), and attention flexibility (d = 0.61). Effect sizes were also large for differences between groups for symptom measures in three of the RCTs. Further, the authors called for more evaluations of ATT against comparable interventions, such as mindfulness based interventions, including follow-up intervals (Knowles et al., 2016).

A general goal of ATT is to increase attention flexibility, often referred to as part of or similar to the concept of attentional control (Callinan et al., 2015). Attentional control can be described as the ability to direct and control attention voluntarily (Derryberry and Reed, 2002). A study suggested that poor attentional control limits the ability of emotion regulation, whereas high attentional control allows the individual to more flexibly disengage and orient attention away from threatening information (Derryberry and Reed, 2002). Hence, poor control may leave the individual vulnerable to emotional disorder. Attentional control is also negatively correlated to state anxiety (Spada et al., 2010) and a possible moderator of the relationship between activation of the CAS and symptoms of emotional disorder (Fergus et al., 2012). It has been demonstrated that ATT can strengthen attentional control measured by self-reported attention flexibility, with medium to large between-group effect sizes (range: η_p² = 0.12–0.15) (Nassif and Wells, 2014; Callinan et al., 2015). This was consistent with performance on a more objective laboratory-based task of attentional control (Callinan et al., 2015). It has thus been suggested that attentional flexibility/control might be a transdiagnostic protective factor and a putative change mechanism of ATT (Fergus and Bardeen, 2016).

Originating in Buddhist traditions, mindfulness can be defined as “the awareness that emerges through paying attention on purpose, in the present moment, and non-judgmentally to the unfolding of experience moment by moment” (Kabat-Zinn, 2003, p. 145). Thus, mindfulness can be regarded as consisting of two components: self-regulation of attention toward current experiences, and relating to these experiences in an open, curious, and accepting stance (Bishop et al., 2004). Attentional processes are important in mindfulness, such as focusing on inner experiences of breathing and emotional sensations. Research also indicates that mindfulness training has the potential to modify and strengthen attention following regular training, such as enhancing the ability to voluntarily shift focus of attention (Jha et al., 2007; Zylowska et al., 2008; Hölzel et al., 2011).

There is increasing support for the beneficial effects of mindfulness-based interventions and treatments. Different programs have been developed and evaluated, such as mindfulness-based cognitive therapy (MBCT; e.g., Segal et al., 2002) and mindfulness-based stress reduction (MBSR; e.g., Kabat-Zinn, 1982, 1990). A meta-analysis including 39 studies demonstrated that MBCT, MBSR, or similar interventions were effective in reducing symptoms of depression and anxiety in clinical and non-clinical samples, with medium to large effect sizes (Hofmann et al., 2010).

Based on research indicating that self-compassion might be one of several key mechanisms accounting for the positive effects following mindfulness-interventions, a MSC-program has been developed (Germer and Neff, 2013; Neff and Germer, 2013). According to Neff (2003b), self-compassion consists of three interrelated components: self-kindness, a sense of common humanity, and mindfulness. In these terms, being self-compassionate means relating to oneself in a friendly and patient manner, understanding that pain and suffering is experienced by all humans, and being mindfully aware of painful experiences without over-identifying with them. Mindfulness in the context of self-compassion is described as awareness in a balanced way to one's negative thoughts and emotions, and is thus slightly more specific than mindfulness in general (Neff, 2003b). The MSC-program is originally designed as an intervention of eight weekly group meetings, for both clinical and non-clinical populations, aiming at enhancing self-compassion through informal (during daily life) and formal (sitting meditation) exercises (Germer and Neff, 2013; Neff and Germer, 2013).

Research indicates that self-compassion is inversely related to psychopathology (Barnard and Curry, 2011), and a meta-analysis summarizing 20 cross-sectional studies found large effect sizes for the negative relationship between self-compassion and stress, anxiety, and depression (MacBeth and Gumley, 2012). Altogether, this indicates that self-compassion might increase resilience against stress and be an important buffer against psychopathology. One RCT evaluating the MSC-program with a non-clinical sample, found the program to be effective compared to a waitlist control (Neff and Germer, 2013). Between-group effect sizes were large for self-compassion and depression, small for stress, and medium for remaining measures such as anxiety and mindfulness. Another study of particular relevance for the current study found promising results with a briefer self-compassion intervention with a non-clinical sample of 52 students (Smeets et al., 2014). The RCT compared a self-compassion program of three weekly meetings to a time-management control group and demonstrated large effect size for self-compassion (d = 1.19), medium effect sizes for optimism (d = 0.66), self-efficacy (d = 0.52), and reduction in rumination (d = 0.70), and small effect size for worry (d = 0.19). This indicates that a three-session trial may be sufficient for a therapeutic effect, such as improving well-being and resilience (Smeets et al., 2014). However, this study did not include follow-up assessment.

As presented above, ATT and mindfulness originate from different traditions and as such have several dissimilarities. Meditation for instance, which is a core element of mindfulness-based interventions, is not recommended in MCT (Wells, 2009). Furthermore, although attention is emphasized in both mindfulness and ATT, they seem to differ in their perspectives on the preferential locus or direction of attentional focus. Self-focused attention can be defined as: “An awareness of self-referent, internally generated information that stands in contrast to an awareness of externally generated information derived through sensory receptors” (Ingram, 1990, p. 156). Heightened self-focused attention has traditionally been associated with psychopathology, and is considered a core component shared by several psychological disorders, such as anxiety and depression (Ingram, 1990). ATT targets inflexible and excessive self-focused attention, aiming to increase attention flexibility, and switch to a more external attentional focus (e.g., Wells, 2009). In the mindfulness tradition, however, increased internal attentional focus has been suggested as an important change mechanism for achieving the beneficial effects of mindfulness training (Baer, 2009). Distinct functions of different types of self-focused attention has been proposed (Trapnell and Campbell, 1999), and a ruminative, self-critical self-focus, as described in MCT, is probably different from the reflective, experiential self-focus associated with the mindfulness tradition (Baer, 2009). Studies have also supported this notion (Watkins and Teasdale, 2004).

Despite the conceptual differences, both intervention perspectives seem promising in reducing symptoms of anxiety and depression, and may operate through related mechanisms such as attentional processes. Therefore, the relationship between them are of interest to explore. This was done in a RCT with 76 students, comparing ATT to mindfulness-based progressive muscle relaxation (MB-PMR) in a single-session trial (Fergus et al., 2014). Symptoms of cognitive and somatic anxiety were significantly reduced after one session, with medium to large effect sizes in both the ATT-group (range: d = 0.32–0.65) and the MB-PMR-group (range: d = 0.59–1.04)¹. Heightened self-focused attention was related to less anxiety after MB-PMR, whereas heightened externally focused attention was related to less anxiety after ATT. However, the study was a single-session trail and lacked follow-up assessment. Despite its limitations, this RCT indicates that both perspectives are effective in reducing symptoms of anxiety, and that the effect of self-focused attention might depend on whether it is performed in a mindfulness-based context or not (Fergus et al., 2014).

The effectiveness of these perspectives in reducing anxiety symptoms via common processes has also been supported in a recent RCT comparing ATT and MB-PMR to a thought wandering control (TWC) with 81 high trait anxious individuals (McEvoy et al., 2017). There was a significant reduction in state anxiety after a single session, with large effect sizes in both experimental groups (ATT: d = 0.83; MB-PMR: d = 0.85). Contrary to previous findings (Fergus et al., 2014), the notion that internal vs. external shifts in attention is associated with symptom reduction was not supported in this study. Furthermore, cognitive flexibility as measured by an emotional Stroop task was not associated with anxiety reductions. This was inconsistent with their hypothesis, previous studies (e.g., Nassif and Wells, 2014) and metacognitive theory (Wells, 2009). However, changes in present-focused attention and metacognitive beliefs were potent change mechanisms across ATT and MB-PMR. The authors concluded that the two techniques are more similar than different and may influence symptom reduction via common mechanisms (McEvoy et al., 2017). A related study exploring the theoretical basis of such interventions also found that mindfulness and metacognitions share important elements, although they are distinguishable constructs (Solem et al., 2015).

In summary, metacognitive and mindfulness-based traditions offer viable treatment options and may contribute to increased understanding of emotional disorder through different theoretical perspectives on self-regulation and attentional processes (e.g., Hofmann et al., 2010; Normann et al., 2014). Specifically, both ATT and MSC can be considered promising cost effective and easy-to-administer interventions for preventing and reducing symptoms of anxiety and depression (Neff and Germer, 2013; Knowles et al., 2016). Such interventions are originally built upon separate constructs, but may be somewhat overlapping (Solem et al., 2015; McEvoy et al., 2017). Thus, comparing these perspectives is of interest, as well as exploring how underlying mechanisms such as attentional control relates to symptom reduction (Fergus and Bardeen, 2016). As few controlled studies have evaluated these interventions, and only two studies have compared ATT and a mindfulness-based intervention (Fergus et al., 2014; McEvoy et al., 2017), more RCTs are needed.

The current RCT therefore sets out to compare the efficacy of ATT and MSC in a three-session trial over 3 weeks, thus expanding results from Smeets et al. (2014), Fergus et al. (2014), and McEvoy et al. (2017). However, the current RCT aims at overcoming limitations in these studies. Fergus et al. (2014) and McEvoy et al. (2017) did not include measures of depression or mindfulness and Smeets et al. (2014) did not include measures of anxiety and depression. Therefore, measures of both anxiety and depression will be included in addition to theoretical construct measures. Due to lacking information about long-term effects in the previous studies, follow-up assessment will also be included. The aim of the current RCT is to test the following hypotheses:

H1: Both interventions will lead to a significant reduction in symptoms of anxiety and depression. H2: Both interventions will give a significant increase in mindfulness, attention flexibility, and self-compassion and treatment-responders will experience more change than non-responders on these measures.

A total of 94 participants showed interest in participation, of which 81 showed up to intervention. The total sample therefore consisted of 81 Norwegian undergraduate and graduate students at the Norwegian University of Science and Technology (NTNU), mean age 22.9 (SD = 3.3, range = 18–36). Participation was open for everyone interested. The participants were not assessed with diagnostic interviews, but had self-reported symptoms of depression, anxiety, and stress. The majority of the participants were female (75.3%) and 55.6% reported having a partner. In total, 69 participants completed the three-session intervention. A total of nine participants (22.0%) dropped out in the MSC condition and three participants (7.5%) in ATT, no reasons were reported. Participant flow is presented in Figure 1.

The study was a RCT approved by the Regional Medical Ethics Committee in Norway (ref.nr. 2015/470). Informed written consent was given from all participants. Intervention was implemented from 2015 to 2017. Participants were recruited at two NTNU campuses with flyers, posters, and promotion in lectures and social media. The study was presented as a course in stress management based on either mindfulness or attention training consisting of three group meetings. Participation was open for everyone interested, but the information implied that the course was suited for people experiencing excessive stress and worry. Using the Research Randomize Program (www.randomizer.org), the recruited participants were randomized to either the ATT or MSC experimental group. Participants were blind as to which experimental group they were allocated until the first group meeting. All participants completed an online questionnaire (described below) before, 1 week after, and at 6-month follow-up.

The questionnaire consisted of demographics and three 1-item questions about level of test anxiety, self-esteem, and loneliness using a 4-point scale; general symptom measures of anxiety and depression used as primary outcome measures; and treatment-specific measures for evaluation of the constructs of mindfulness, self-compassion, and attention flexibility.

The intervention consisted of three group sessions 3 weeks in a row, of either ATT or MSC, as well as instructions to listen to certain audiotapes every day between sessions during the intervention period. Each group consisted of 6–10 participants sitting in a circle, being co-led by two therapists. All group sessions were held in the afternoon at the university campus and lasted for 45 min. The participants were instructed to take an active part in the group discussions while the therapists mainly were facilitating, asking socializing questions, and unraveling misunderstandings.

The following structure was used in the first two sessions in all groups:

Agenda-setting for the day.

Session two and three in both conditions began by asking about homework, feedback from the participants, summing up the rationale, and unraveling possible misunderstandings. In the second session, this was followed by step three to four, as presented above.

In the third and last session, there was no practicing with audiotapes. Most of the time went to group discussion on the principles the participants had learned during the 3-week intervention. Each participant described in turn how they had experienced listening to the audiotapes and how they could relate the principles to their everyday life. Finally, participants provided evaluation feedback of the course. Differences between the conditions are described below.

The therapists were four clinical psychology students on their fifth year, under supervision from a clinical psychologist. The student therapists had no prior official training in ATT or MSC. Training involved extensive literature reading on the two conditions and receiving feedback on videotaped recordings of training sessions before the experimental group interventions started. Therapists also discussed and trained with the audiotapes themselves. All therapists conducted both ATT and MSC equally. The group leaders were supervised between the group sessions.

To compare the two samples on demographics and measures pre-intervention, one-way analysis of variance (ANOVA) was used. Chi-square tests were used to compare the groups with respect to dichotomous variables.

Repeated measures ANOVA was used to assess changes in measures from pre- to post-intervention and follow-up, and a split-plot ANOVA was used to compare the groups with respect to changes in outcome measures. Pre-intervention, 2.5% values were missing on GAD-7, FFMQ, DMQ, and SCS-SF. Missing data in these measures were replaced by mean values. Last observation carried forward was used for missing data post-intervention for dropouts (7.5% for ATT and 22.0% for MSC) and missing follow-up data. To assess whether homework frequency influenced outcome, a split-plot ANOVA with number of homework exercises as a covariate was conducted. Missing data on the homework-variable was not replaced. Cohen's d with pooled SD and partial eta squared was calculated and reported as effect sizes. Cohen's d was calculated for each measure for each group, and correlations between pre- and post-intervention values were included in the calculation. This was done using Morris and DeShon's (2002) Equation (8), which corrects for dependence between means. Cohen's d is interpreted as small (0.2), moderate (0.5), and large (0.8) effect size (Cohen, 1988). For main effects and differences between groups, partial eta squared was used as measure of effect size. Partial eta squared is interpreted as small (0.01), medium (0.06), and large (0.14) effect size (Richardson, 2011).

It was of interest to examine differences between responders and non-responders to intervention in order to find out for whom the intervention worked and why. Response to intervention was defined as at least 35.0% improvement in primary outcome symptom measures (PHQ-9 and GAD-7) pre- to post-intervention. ANOVAs were run to compare non-responders and responders in the ATT- and MSC-group on change scores in mindfulness, self-compassion, and attention flexibility. For an easier to interpret graphic presentation, individual change scores on all theoretical measures were rescaled to a standardized 0–100 scale.

Finally, linear regression analyses were conducted using the total sample to predict primary outcome measures post-intervention, using primary outcome measures pre-intervention, age, gender, and change in mindfulness, self-compassion, and attention flexibility as predictors. This was done in order to determine which of these variables contributed to treatment response across conditions.

An overview of demographics, symptoms, and other measures in the two experimental groups pre-intervention is presented in Table 1. One-way ANOVA indicated no significant differences between groups on any measures pre-intervention. None of the participants reported no symptoms (scores of 0 on all measures). Categorizing symptoms into none, mild, and moderate to severe symptoms, 37.0% showed no symptoms of depression, while 63.0% scored in the mild to severe range. As for symptoms of anxiety, 35.8% showed no symptoms, while 64.2% scored in the mild to severe range. A total of 39.0% reported being a little to very lonely, 21.9% reported having a little bad to bad self-esteem, and 50.6% reported having some to a lot test anxiety.

Repeated measures ANOVAs indicated a significant reduction in depressive symptoms (PHQ-9) and anxiety symptoms (GAD-7) in both groups pre- to post-intervention, with medium effect sizes (range: d = 0.53–0.71) and no significant differences between conditions (see Table 2). These results were stable, with no significant differences in symptom level between post-intervention and follow-up either for PHQ-9 (p = 0.872) or GAD-7 (p = 0.934). As presented in Table 3, participants reporting no symptoms increased in both groups for PHQ-9 (ATT = 40.0–55.0%; MSC = 34.1–53.7%) and GAD-7 (ATT = 37.5–60.0%; MSC = 34.1–51.2%), pre- to post-intervention. There was also a substantial decrease in participants scoring within the moderate to severe range. In general, these results were maintained at follow-up, as depicted in Table 3. However, there was a slight increase in participants scoring within the mild range. Defining treatment response by minimum 35.0% reduction in symptoms measures, there was a treatment response post-intervention of 33.3% (GAD-7) and 32.1% (PHQ-9) across conditions. This was maintained at follow-up (GAD-7 = 37.0%; PHQ-9 = 35.8%).

As presented in Table 2, there was a significant increase in self-compassion (SCS-SF), attention flexibility (DMQ flexibility), and mindfulness (FFMQ) in both groups pre- to post-intervention, with large effect sizes and no significant differences between groups. With small to medium effect sizes, there was also a significant increase in self-esteem and a significant decrease in test anxiety, with no significant differences between groups. For self-compassion, attention-flexibility, and self-esteem, these results were maintained with no significant differences between post-intervention and follow-up (SCS-SF: p = 0.356; DMQ flexibility: p = 0.618; self-esteem: p = 0.358). For mindfulness, there was a significant increase (FFMQ: p = 0.010) and for test-anxiety there was a significant decrease (p = 0.003) from post-intervention to follow-up. The replacement of missing follow-up data did not affect the results of the analysis as opposed to analyses with completers only.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.