Association Between Dispositional Mindfulness, Clinical Characteristics, and Emotion Regulation in Women Entering Substance Use Disorder Treatment: an fMRI Study

Dispositional mindfulness (DM) is associated with emotion regulation (ER) in healthy populations and may be protective for individuals with substance use disorders (SUD). We tested hypotheses concerning the associations of DM with ER, mental health symptoms, drug use severity, and behavioral and brain metabolic responses during an emotional Go-Nogo task. Women entering an SUD treatment program (N = 245) self-reported on the Five Facet Mindfulness Questionnaire (FFMQ); Depression, Anxiety, and Stress Scale (DASS-21); Addiction Severity Index (ASI); and Difficulties in Emotion Regulation Scale (DERS). A subgroup of 45 women completed the emotional Go-Nogo task while undergoing fMRI. Associations between DM and self-reported ER and clinical characteristics were tested in the full sample. Associations between DM and behavioral and neural responses during the Go-Nogo emotion regulation challenge were tested in the fMRI sub-sample. In the full sample, FFMQ correlated with DASS-stress (r =  − .43, p < .0001), DASS-depression (r =  − .52, p < .0001), DASS-anxiety (r =  − .32, p < .0001), DERS (r =  − .66, p < .0001), and ASI-Drug scores (r =  − .28, p < .0001). In the subsample, inhibition of the natural avoidance response while approaching a fearful stimulus recruited activity in the inferior frontal gyrus and the insular cortex (IC). Activity in the IC cluster was positively correlated with FFMQ scores (r = 0.38, p = .015). FFMQ scores correlated with activation in the striatum and IC during face processing. Among women with SUD, DM beneficially correlated with ER, mental-health symptoms, addiction severity, and recruitment of neural substrates underlying ER.

1995) and social functioning (Phillips et al., 2008). Adaptive emotion regulation strategies are especially important for people with substance use disorders (SUD), who often use drugs or alcohol to cope with challenging emotions (Bolton et al., 2009). Emotion dysregulation has been identified as a risk and causal factor in the initiation, development, and maintenance of SUDs (Kober, 2014), and negative affect is furthermore known to be a primary trigger for relapse (Witkiewitz & Villarroel, 2009). In general, individuals with SUDs demonstrate weaker emotion regulation skills compared to those without SUDs (Fox et al., 2008). However, improvements in emotion regulation following SUD treatment have predicted reduced substance use among women (Axelrod et al., 2011).
The most studied emotion regulation strategy is reappraisal, which entails reinterpreting the meaning of emotional stimuli in a manner that reduces their negative emotional impact (Martelli et al., 2018;Ochsner & Gross, 2008). For example, when suffering a personal rejection, a person may reframe the situation as "when one door closes, another one opens." Or more typical of an experimental setting in which participants view an image of a horrific scene, the viewer can reappraise it as, for example, a movie set with actors. This type of emotion regulation appears to be dependent on the prefrontal cortex (PFC) and anterior cingulate, both of which contribute to the modulation of activity in subcortical regions (e.g., amygdala, ventral striatum) involved in affective processing (Ochsner et al., 2012). However, evidence suggests that SUD is characterized by structural and functional deficits in the PFC, which is hypothesized to be at least partially responsible for emotion dysregulation in this population (Kober, 2014).
Emotion regulation has been both conceptually and empirically linked to mindfulness (Guendelman et al., 2017;Hölzel et al., 2011). Mindfulness is defined as nonjudgmental awareness of the present moment in which thoughts, feelings, and sensations are recognized and accepted as they arise (Bishop et al., 2004). It has been suggested that mindfulness relates to implicit rather than explicit emotion regulation (Guendelman et al., 2017). Implicit regulation is characterized by lower emotional reactivity, implying reduced activation in the limbic circuits, thereby reducing the need for effortful top-down cognitive control and inhibition (Brown et al., 2013;Martelli et al., 2018) involved in cognitive reappraisal. Given extensive evidence of compromised PFC functioning, particularly during early abstinence (Goldstein & Volkow, 2002), reduced need for PFC engagement during emotion regulation may be particularly beneficial for SUD patients. Thus, dispositional mindfulness may be an especially important characteristic for this population because it may relate to reduced limbic reactivity to emotional stimuli. The positive association between dispositional mindfulness and emotion regulation has been repeatedly demonstrated in healthy populations (Brown et al., 2013;Droutman et al., 2019Feldman et al., 2007;Ortner et al., 2007). Also, initial evidence suggests that a mindfulness-based adjunct to women's SUD treatment improves both treatment success (days abstinent) and emotion regulation (Price, et al., 2019a(Price, et al., , 2019b. However, little research exists on the association between dispositional mindfulness and emotion regulation in the context of SUD. It is furthermore unknown to what degree DM is associated with brain responses during a task that engages emotion regulatory systems in this population. A negative association between dispositional mindfulness and measures of emotional disturbance (depression, anxiety, and stress) has been identified in healthy participants (Brown et al., 2007;Feldman et al., 2007). Also, greater baseline dispositional mindfulness has predicted less severe depressive symptoms and lower volatility of negative emotional affect in abstinent smokers during a quit attempt (Adams et al., 2014). Thus, we hypothesized dispositional mindfulness would be a protective factor for SUD patients such that higher dispositional mindfulness would be associated with lower self-reported anxiety, depression, stress, and substance use and with better emotion regulation.
In the Go-Nogo task (Hare et al., 2005), the participant is presented with a rapid flow of emotional faces and asked to press a response button as quickly as possible each time the facial expression matches a particular target emotion (which is specified at the beginning of the block) and not respond to other nontarget emotion categories. Because positively and negatively valenced emotion cues create rapid automatic approach and withdrawal response tendencies (respectively), the task engages emotion regulatory processes. First, pressing a button when confronted with a fearful face requires overcoming a natural withdrawal response. Prior work shows that participants take more time on such trials (Hare et al., 2005). The increase in response time constitutes a measure of emotion regulation such that a smaller increase in reaction time is interpreted as better emotion regulation. Second, avoiding a button press when confronted with a happy face requires overcoming an automatic approach response. Prior work showed that participants make more false alarm errors when the nontarget emotion is happy (Hare et al., 2005). We hypothesized that dispositional mindfulness would negatively correlate with response time on trials where the target emotion was fear and with false alarms on trials where the non-target emotion was happy.
Unlike reappraisal, which involves explicit cognitive effort to reassess the event that triggered the emotion (Strauss et al., 2016), mindfulness-based emotion regulation involves redirecting attention to the physiological sensations resulting from the emotion, thereby emphasizing interoceptive awareness (Chambers et al., 2009), and equanimous observation of physiological reactions to a provocative stimulus (Guendelman et al., 2017). Thus, the neurobiological focus of this investigation was the insular cortex (IC), the core neural substrate underlying interoceptive awareness (Craig, 2003;Naqvi & Bechara, 2010) that is also involved in emotion processing (Craig, 2004;Uddin et al., 2017), decision-making (Droutman et al., 2015a, b), and addiction (Droutman et al., 2015a, b).
Structural neuroimaging studies of long-term meditators, who as a group demonstrated enhanced emotion regulation capacity (Lykins & Baer, 2009), show increased cortical thickness in the IC (Lazar et al., 2005). Increased engagement of the IC during experiential focus (Farb et al., 2007) and interoceptive awareness exercises (Farb et al., 2013) was observed in participants following mindfulness training as compared to controls. Mindfulness-based emotion regulation directs attentional resources toward the limbic pathway (including the IC) for present-moment awareness instead of focusing on evaluative processing supported by the medial prefrontal cortex (Farb et al., 2012). This approach of limiting cognitive elaboration in favor of moment-by-moment awareness appears to reduce automatic negative self-evaluation and increase tolerance for negative affect and pain, which can be particularly helpful for chronically dysphoric individuals (Farb et al., 2012) and individuals with SUDs (Price et al., 2019a(Price et al., , 2019b. Thus, identifying an association between IC engagement and dispositional mindfulness was of a particular interest. Data for this study were obtained during the baseline period (prerandomization) from a subset of patients participating in a clinical trial of Moment-by-Moment in Women's Recovery (Amaro & Black, 2021;Vallejo & Amaro, 2009). In the current analysis, we evaluated associations of dispositional mindfulness with clinical characteristics, and with brain activity while participants performed the emotional Go-Nogo task (Hare et al., 2005). We tested the following three hypotheses. First, dispositional mindfulness would positively correlate with self-reported measure of emotion regulation and negatively correlate with self-reported anxiety, depression, stress, and substance use (hypothesis 1). Second, during emotional Go-Nogo task, response time on trials where the target emotion was fear (hypothesis 2a) and false alarms on trials where the non-target emotion was happy (hypothesis 2b) would negatively correlate with dispositional mindfulness. Third, dispositional mindfulness would positively correlate with IC recruitment during emotion processing and regulation (hypothesis 3a) and that IC activation would positively relate to successful emotion regulation (hypothesis 3b).

Participants
Participants were adult women aged 18 to 65 years clinically diagnosed with SUD, admitted to the residential SUD treatment program, fluent in English, and agreed to participate in the parent study (Amaro & Black, 2017). Exclusion criteria were inability to comprehend or sign informed consent, cognitive impairment, an untreated severe chronic mental health condition or untreated psychotic disorder based on clinical intake DSM-IV-TR or DSM-5, older than 65 years of age because this was unusual at the site, reported suicidality during the prior 30 days based on clinical intake assessment, current prisoner, more than 6 months pregnant, and not willing to sign a HIPAA form or be audio recorded during interviews and intervention sessions.
An a priori power analysis was conducted for the parent study and estimated the needed sample size of 206 to detect a between group effect (d = 0.5) of linear trend at 8-month follow-up. For further details on recruitment process and power analysis, see the parent study (Amaro & Black, 2017). As shown in Table 1, the sample was comprised of 245 women with diverse race and ethnicity, characterized by significant socioeconomic disadvantage and varying degrees of exposure to traumatic events and mental health symptomatology. Most participants (91%) reported use of methamphetamines or amphetamines during the 8 months prior to treatment entry. The majority of our sample were multi-substance users, characteristic often reported for methamphetamine users Ling et al., 2014;McKetin et al., 2013). Other most reported substances were marijuana, alcohol to intoxication, and cocaine (Table 1).
All women who had agreed to be part of the parent study and had completed the baseline interview were screened for the specific fMRI study inclusion criteria (see below). If they met the criteria, informed consent was administered and if women agreed to be part of the study, they were scheduled for the fMRI (see Figure S1 in Supplemental Materials). The following inclusion criteria were used for the fMRI study: clinical trial participants who were right-handed, in good general health, and younger than 50 were further considered for the neuroimaging study. Standard exclusion criteria for the fMRI study were used to insure participants' safety and compliance with the imaging center policy: currently or possibly pregnant; using medical devices (cardiac pacemaker, implanted cardiac defibrillator, etc.); metal fragments including shrapnel or other nonremovable metal devices like dental braces or retainers or an intrauterine device; history of head trauma resulting in loss of consciousness for more than 5 min; documented or subjectively reported claustrophobia; hair extensions or a wig connected by wire; permanent eyeliner; and body mass index greater than 36 (due to the narrow circumference of the scanner bore). A subset of 45 women participated in the fMRI component of the study. Data from three participants were excluded from Go-Nogo task analyses (one due to a panic attack in the scanner and two due to misunderstanding or inability to follow the task instructions). Data from two additional participants were excluded from imaging analyses because their head motion during the imaging data collection exceeded a prior selected threshold of 3 mm.

Procedure
All assessments were conducted prior to entering a randomized controlled trial of Moment-by-Moment in Women's Recovery (Amaro & Black, 2017. The study interviewer made appointments with prospective participants, conducted the informed consent and HIPAA process, and administered the baseline assessment interview. On the day of the imaging assessment, the 45 neuroimaging participants were trained on the tasks prior to the 60-min scanning session. The session included two runs of the emotional Go-Nogo task and structural and resting state acquisitions, along with a second task reported elsewhere (Poppa et al., 2019).

Emotional Go-Nogo Task
We utilized a version of the Go-Nogo task with emotional faces as stimuli, based on Hare et al. (2005). During the task, the participant sees a word identifying an emotion (happy, fear, or calm), followed by a set of faces, and is asked to press a button ("as quickly and as accurately as you can") when the facial expression matches the emotion identified by the word preceding the set (Go, target emotion), but to do nothing for any other emotion (Nogo, non-target emotion) (see Fig. 1). The task consists of four blocks of 32 trials each, where 75% of the trials were "Go-trials" for all subjects in each block. The order of blocks was pseudo-randomized (each participant was randomly assigned one of the four possible order combinations). The image set consists of 12 female faces (five Caucasian, three African American, and four Asian). Models 3,5,6,8,9,11,12,13,14,15,16, and 18 from the NimStim set were used (Tottenham et al., 2009). Faces were presented for a maximum of 500 ms, with fixation cross displayed between trials. Total time of each session was 4 min 48 s. It consisted of four 50 volumes/100 s blocks. Each block included thirty-two 3.125-s trials. Before each block, the emotion name ("FEAR" or "HAPPY" or "CALM") was displayed for 2 s. The task sequence was optimized using optseq2 (Dale, 1999), which creates a schedule for rapid-presentation event-related (RPER) fMRI experiments with jittered onset times for the events, thus removing the overlap from the estimate of the hemodynamic response Participants were asked to press a key for the faces expressing the Go-emotion and do nothing for other emotional faces. In the beginning of each block, emotion name ("FEAR" or "HAPPY" or "CALM") was displayed for 2 s. The word was then replaced by a fixation cross for a jittered period of 2-5 s (mean 2.63 s). Following fixation, a picture was shown for 0.5 s, followed again by intertrial jittered interval of 2-5 s (mean 2.63 s) that otherwise would result from presenting events close in time.

Five Facet Mindfulness Questionnaire
The short form of the Five Facet Mindfulness Questionnaire (FFMQ-SF; Bohlmeijer et al., 2011) measures the general propensity of five distinct, but related, facets of mindfulness with 24 items that are scored as a sum: (a) Observing, defined as noticing or attending to internal and external experiences; (b) describing, defined as labeling internal experiences with words; (c) acting with awareness, defined as attending to one's activities of the moment (as opposite of acting on automatic pilot); (d) non-judging of inner experience, defined as taking a non-evaluative stance toward thoughts and feelings; and (e) non-reactivity to inner experience, defined as allowing thoughts and feelings to come and go, without getting caught up in or carried away by them.
Items are rated on a 5-point Likert scale, ranging from 1 (never true) to 5 (very often true). Cronbach's alpha for the FFMQ in the current sample was 0.80, Mc Donald's omega was 0.86.

Difficulties in Emotion Regulation Scale
The Difficulties in Emotion Regulation Scale (DERS) was developed as a comprehensive measure to evaluate emotion regulation "involving not just the modulation of emotional arousal, but also the awareness, understanding, and acceptance of emotions, and the ability to act in desired ways regardless of emotional state" (Gratz & Roemer, 2004, p. 41). It consists of 36 items in six subscales: (a) Nonacceptance of emotional responses, (b) lack of emotional awareness, (c) lack of emotional clarity, (d) impulse control difficulties, (e) limited access to emotion regulation strategies, and (f) difficulties engaging in goal-directed behavior. The DERS is scored as global sum of scores. Cronbach's alpha for the DERS in the current sample was 0.96, and Mc Donald's omega was 0.97.

Addiction Severity Index
Severity of drug use for the 30 days prior to treatment entry was assessed using the Addiction Severity Index (ASI Drug; McLellan et al., 1992).

Depression, Anxiety, and Stress Scale
The Depression, Anxiety, and Stress Scale (DASS-21; Antony et al., 1998;Henry & Crawford, 2005) is a shorter version of the 42-item DASS questionnaire. The scale has 7 items for each of the three subscales: (a) Depression; (b) Anxiety; and (c) Stress. Cronbach's alpha for the DASS-21 in the current sample was 0.94, and Mc Donald's omega was 0.95.

Distress Tolerance Scale
The Distress Tolerance Scale (DTS; Simons & Gaher, 2005) is a 15-item self-report questionnaire that measures general emotional distress tolerance. Besides the overall factor, the scale has four subscales: (a) Tolerance, (b) Absorption, (c) Appraisal, and (d) Regulation. Participants rated the items on a scale ranging from strongly agree (1) to strongly disagree (5). The total score was used as a measure of distress tolerance, in accordance with Vujanovic et al. (2016). Cronbach's alpha of the DTS in the present sample was 0.88, and Mc Donald's omega was 0.92.

Perceived Stress Scale
The Perceived Stress Scale (PSS-10; Cohen et al., 1983) is a 10-item scale that measures the degree to which an individual appraises situations as stressful. Items assess how unpredictable, uncontrollable, and overloaded respondents find their lives to be, along with current levels of experienced stress. Cronbach's alpha of the PSS in the present sample was 0.85, and Mc Donald's omega was 0.89.

Data Analyses
To test hypothesis 1, that dispositional mindfulness has an inverse relationship with symptom severity for substance use, stress, depression, and anxiety and is positively related to emotion regulation, we calculated Spearman's correlations of behavioral measures obtained at baseline from the full sample of 245 participants. We used Spearman's correlation because dispositional mindfulness measure did not satisfy the Shapiro-Wilk normality test. The Bonferroni method was used to correct for multiple comparisons. Given six correlations, the corrected alpha significance threshold was 0.008. Three behavioral outcome measures were computed based on Go-Nogo task (42 participants): hit rate, which is the proportion of button presses in response to the target (Go) stimulus; false alarm rate, which is the proportion of button presses to the non-target (Nogo) stimulus; and response time (the average response time for correct responses to a Go stimulus). We then examined whether hits, false alarms, and response times differed for emotionally congruent (Go-Happy and Nogo-Fear) and incongruent (Nogo-Happy and Go-Fear) conditions, as expected in this task. To do so, we fitted repeated-measures generalized linear models (GLM) for each outcome (hit rate, false alarm rate, and response time) with emotion as within subjects factor, followed by post hoc pairwise t-tests for significant main effects to confirm expected differences between congruent and incongruent conditions (Hare et al., 2005).
To test hypothesis 2, we correlated FFMQ scores with the difference in the response times and with the difference in the false alarms between congruent and incongruent conditions, specifically, the response time difference between Go-Fear and Go-Happy trials (hypothesis 2a) and the difference in the number of false alarms between Nogo-Happy and Nogo-Fear trials (hypothesis 2b).

Functional Imaging Procedure and Data Acquisition
Participants laid supine in the scanner and viewed visual stimuli back-projected onto a screen through a mirror attached to the head coil. Stimuli were presented by projection screen with native pixel resolution 1024 × 768 and a head coil mounted mirror, with an 88-cm viewing distance subtended 6.5° of visual angle vertical and 5.2° horizontal. Participants' visual acuity was normal or corrected to normal, using contact lens or MR compatible glasses. Foam pads were utilized to minimize head motion. Stimulus presentation and timing of all stimuli and response events were controlled by Matlab R2017a (The Mathworks, Inc.) programs based on Psychtoolbox-3 (http:// www. psych toolb ox. org) extensions on a MacBook Pro. Participant's responses were collected online using an MRI-compatible button box. Images were acquired with a 3-T Siemens MAGNETON Prisma System with a 20-channel head coil. Functional images were obtained using a gradient echo, echo-planar, T2*-weighted pulse sequence (TR = 2,000 ms, one shot per repetition, TE = 25 ms, flip angle = 90°, 64 × 64 matrix size with a resolution of 3 mm 2 ). Forty-one 3-mm axial slices were used to cover the whole cerebrum and most of the cerebellum with no gap.

fMRI Data Preprocessing and Statistical Analyses
Image preprocessing and statistical analysis were carried out using the FMRI Expert Analysis Tool (FEAT), version 6.00, part of the FSL package in the FMRIB software library, version 4.1.8 (www. fmrib. ox. ac. uk/ fsl). The data were temporally filtered using a nonlinear high-pass filter with a 100-s cutoff and spatially smoothed using a 5-mm full-width-half-maximum (FWHM) Gaussian kernel. A twostep registration procedure was used whereby images were first registered to the MPRAGE structural image and then into the standard Montreal Neurological Institute MNI-152 T1 template brain, using affine transformations with FLIRT (Jenkinson & Smith, 2001;Jenkinson et al., 2002). Registration from MPRAGE structural images to standard space was further refined using FNIRT nonlinear registration (Andersson et al., 2007). Statistical analyses were performed in the native image space, with the statistical maps normalized to the standard space prior to higher-level analysis. Multivariate Exploratory Linear Optimized Decomposition into Independent Components (MELODIC; ) was used to denoise the preprocessed functional data. The FIX software package was used to automatically identify noise components Salimi-Khorshidi et al., 2014). Data were modeled at the first level using a general linear model within FSL's FILM module. All group analyses were performed using a random-effects FLAME (FMRIB's Local Analysis of Mixed Effects) stage 1 simple mixed effect model Woolrich, 2008;Woolrich et al., 2004). Group images were then thresholded using cluster detection statistics with a height threshold of z > 2.3 and a cluster probability of p < 0.05, corrected for whole-brain multiple comparisons using Gaussian random field theory.

Analytical Approach
Emotional Go-Nogo is a multifaceted task that examines behavioral, motor, and emotion-related inhibition (emotion regulation) and emotion processing. Here, we focus on measures of emotion-related inhibition or emotion regulation. The historically used approach of contrasting BOLD signal for Nogo and Go trials to examine neural correlates of behavioral and motor inhibition was not included here because it was not a focus of the current inquiry. Furthermore, this contrast has been shown to be problematic (Casey et al., 2001), because most of the activation typically identified using this approach is driven by attention and working memory rather than inhibitory processes (Criaud & Boulinguez, 2013). To examine brain responses associated with inhibiting the natural avoidance response (i.e., downregulation of negative emotion), we contrasted the BOLD response on trials where target emotion was fear (Go-Fear trials) with Go-Happy trials and Nogo-Fear trials. To examine the inhibition of natural approach tendencies to positive emotions (i.e., down-regulation of positive emotion), we contrasted BOLD responses for trials where the non-target emotion is happy (Nogo-Happy) with Nogo-Fear trials and Go-Happy trials. As a side note, although the neutral emotion (CALM condition) is usually used as a basis for comparison, our participants found CALM faces ambiguous. This was evident from the postscan debriefing interviews and slower response times on CALM trials. Notably, during debriefing, many participants reported experiencing the neutral expression as fearful or angry. Hence, the CALM condition was not used for any hypothesis-driven analyses.
Finally, we examined emotion perception by contrasting emotional face processing with the baseline (Berkman et al., 2009;Tottenham et al., 2011) separately for each emotion and by averaging across all emotions.

Whole-brain Analysis 1
In this whole-brain analysis, six predictors of interest (2 trial types (Go, Nogo) × 3 emotions (fear, happy, calm)) were used to model correct trials (button press on Go trials and no response on Nogo trials). Error trials (no button press on Go trial or button press on Nogo trials) from the six conditions were added to the model as separate nuisance covariate. We modeled the main effect of each type of trial for each emotion by contrasting against baseline (inter-trial break periods when fixation cross was displayed), specifically, Go-Fear, Go-Happy, Nogo-Fear, and Nogo-Happy. Also, we modeled down-regulating negative emotion by contrasting Go-Fear trials with Go-Happy trials and with Nogo-Fear trials, and down-regulating positive emotion by contrasting Nogo-Happy trials with Nogo-Fear trials and with Go-Happy trials.
At the group level, we examined the group mean effect of the four contrasts of interest: Go-Fear trials versus Go-Happy trials, Go-Fear trials versus Nogo-Fear trials, Nogo-Happy trials versus Nogo-Fear trials, and Nogo-Happy trials versus Go-Happy trials. We also examined the relationship between FFMQ, our dispositional mindfulness measure, and the main effects for each emotion and trial type (Go-Fear, Nogo-Fear, Go-Happy, Nogo-Happy) by adding demeaned FFMQ as a covariate in the higher-level model.

Whole-brain Analysis 2
The goal of this analysis was to examine the general impact of dispositional mindfulness on processing emotional faces. Three emotion predictors (fear, happy, calm) were used to model all trials (correct and error) without differentiating the action required (combining Go and Nogo trials). We modeled the main effect of each emotion by contrasting against baseline (inter-trial break periods when fixation cross was displayed). At the group level, we modeled the group mean effect of the three main contrasts and, more importantly, to identify mindfulness-related neural substrates in emotion perception, we included FFMQ scores as a covariate in these analyses.

ROI Analysis
Given our hypothesis (hypothesis 3) that IC is an important neural substrate underlying the role of mindfulness in emotion regulation, we performed region-of-interest (ROI) analyses by creating masks of the activation maps derived from the emotion regulation contrasts (that is, contrasts that isolate demands associated with emotion-response incongruence), as discussed in the section on whole-brain analysis 1. These contrasts were Go-Fear versus Go-Happy, Go-Fear versus Nogo-Fear, and Nogo-Happy versus Nogo-Fear, constrained by our IC mask. To isolate the IC, we first created probability masks of the right hemisphere IC components: anterior dorsal (dAIC), anterior ventral (vAIC), and posterior (PIC) (Chang et al., 2012). We then flipped the x-axial using the fslswapdim tool from FSL to generate masks for the left PIC, dAIC, and vAIC; we then combined all 6 components into one IC mask using the fslmaths tool from FSL. Parameter estimates isolated from the ROIs for these contrasts were extracted for each participant, then correlated with the dispositional mindfulness measure (FFMQ).

Down-regulation of Negative Emotion
To isolate the additional processing required to override a fear-elicited bias to withdraw, we conducted a whole-brain analysis contrasting Go-Fear and Go-Happy trials. This contrast identified clusters in the inferior frontal gyrus (IFG) and dorsomedial PFC bilaterally (see Fig. 2A, Table 2). The contrast of Go-Fear and Nogo-Fear trials further identified clusters in right IFG, ventromedial PFC (vmPFC), and left posterior IC (see Fig. 2B, Table 2). Given our hypothesis that the IC is a key neural substrate for mindful emotion regulation (hypothesis 3b), we created an ROI using the IC clusters from Go-Fear > Nogo-Fear contrast (see Fig. 2C). Using this mask, we extracted the mean parameter estimates from the Go-Fear trials and assessed its association with the FFMQ. We observed significant association between this measure of task-related neural activity in the IC and FFMQ score (r = 0.38 p = 0.015) (see Fig. 2D).

Down-regulation of Positive Emotion
Whole-brain analysis contrasting Nogo-Happy and Go-Happy trials identified clusters in the left IFG and dorsal PFC (see Fig. 2E, Table 2). There was no significant difference between Nogo-Happy and Nogo-Fear trials.

Processing Emotional Stimuli
While participants attended to emotional faces on both Go and Nogo trials (collapsed across emotion), FFMQ scores were positively correlated with activation in the caudate, putamen, thalamus, and anterior IC (see Fig. 3A, Table 2). When isolated by emotion, we found that when attending to positive (Fig. 3B) and negative (Fig. 3C) emotional faces during Go-trials, FFMQ scores were positively associated with signals from the precuneus, cingulate cortex, and the IC ( Table 2). The positive correlation of FFMQ and activation in the IC while attending to emotional faces provides support for hypothesis 3a. We found no significant association Fig. 2 Whole-brain analysis, examining down-regulation of negative and positive emotion. Group images thresholded using cluster detection statistics with a height threshold of z > 2.3 and a cluster probability of p < 0.05, corrected for whole-brain multiple comparisons using Gaussian random field theory. A Contrast between Go-Fear and Go-Happy trials. B Contrast between Go-Fear and Nogo-Fear trials, insula region of interest circled. C Insular cortex in yellow; contrast between Go-Fear and Nogo-Fear trials in blue; insular ROI circled. D PE extracted using insula ROI correlated with FFMQ. E Contrast between Nogo-Happy and Go-Happy trials between FFMQ scores and brain activation on Nogo-trials of any emotion when contrasting with baseline.

Discussion
In this study, we tested whether dispositional mindfulness was correlated with emotion regulation, mental health, and addiction severity symptoms and examined behavioral and neural correlates of dispositional mindfulness and its relationship with key clinical constructs in women initiating treatment for SUDs.
Higher dispositional mindfulness was associated with greater emotion regulation ability, lower addiction severity, and fewer reported symptoms of stress, depression, and anxiety (i.e., support for hypothesis 1). A growing literature has demonstrated improvement in stress (Hölzel et al., 2010), depression (Teasdale et al., 2000), and anxiety (Davidson et al., 2003) and reduced substance misuse (Li et al., 2017) following mindfulness-based interventions (MBIs). These interventions are focused on actively cultivating mindful disposition and show increases in mindfulness scores and thus their findings are relevant and consistent with the current research. However, a recent meta-analysis found only Table 2 Whole-brain analysis examining down-regulation of negative and positive emotion, and correlational whole-brain analysis identifying areas where activation is mediated by FFMQ during emotion processing. The value in the "max" column is the maximum z-stat in each cluster. X, Y, and Z are x, y, and z coordinates in MNI-152 space in millimeters. The number of voxels in the cluster is recorded in the "voxels" column partial support for the differential sensitivity of mindfulness questionnaires to changes with treatment (Baer et al., 2019) and thus our findings provide novel support for the associations between dispositional mindfulness and these clinical measures.
We used an emotional Go-Nogo task to assess the association between dispositional mindfulness and both behavioral and neural responses during an emotion regulation challenge. The emotional Go-Nogo task assesses rapid emotion regulation by leveraging automatic response tendencies associated with emotion cue processing (specifically, approach response to happy and withdrawal response to fear stimuli). Consistent with prior work using this paradigm, we observed that the presence of emotional cues impacted response time and accuracy of task performance (Dreyfuss et al., 2014;Lee et al., 2018;Somerville et al., 2011). Specifically, we observed slower responses to fear targets than to other emotion targets, and more false alarm responses to happy targets than to other targets. We found no relationship between these behavioral measures and FFMQ scores (i.e., Fig. 3 Correlational whole-brain analysis identifying areas where activation is mediated by FFMQ during emotion processing. Group images thresholded using cluster detection statistics with a height threshold of z > 2.3 and a cluster probability of p < 0.05, corrected for whole-brain multiple comparisons using Gaussian random field theory. A Activation mediated by FFMQ during processing of FEAR or HAPPY emotional faces (both Go and Nogo trials). B Activation mediated by FFMQ during Go-Happy trials. C Activation mediated by FFMQ during Go-Fear trials null effects for hypothesis 2). It is possible that our subsample size was not large enough to capture this relationship.
In neuroimaging analyses, we examined brain responses associated with inhibiting the natural avoidance response (i.e., down-regulation of negative emotion) by contrasting the BOLD response on Go-Fear trials with Nogo-Fear trials. This allowed us to isolate activation in brain regions associated with overriding the fear-related automatic withdraw response tendency and to engage an approach response. This emotion regulation challenge recruited activity in the IFG and posterior IC. It has been suggested that the IFG is a key neural component responsible for inhibitory control (Aron et al., 2003;Cai et al., 2014). It has also been proposed that the IFG has a special role in reactive behavioral control that incorporates fast associative learning that is adaptive in less predictable environments (Tops & Boksem, 2011). Moreover, the activation in the above identified IC cluster was positively correlated with dispositional mindfulness and hit rate, such that higher dispositional mindfulness corresponded to higher activation in the IC cluster and better behavioral performance (i.e., support for hypothesis 3). Thus, IC activity is relevant to understanding the relations between mindfulness and emotion regulation.
We furthermore evaluated neural substrates of socioemotional processing. That is, we examined participants' brain responses to faces across all emotion categories. We found that dispositional mindfulness was positively correlated with activation in the vmPFC and striatum across all emotions. Moreover, during positive (happy) and negative (fear) emotional processing, dispositional mindfulness was positively correlated with activity in the IC. It has been suggested that the striatum, vmPFC, IC, and amygdala are four key regions involved in generating emotion (Ochsner et al., 2012). The striatum is involved in learning which cues (like smiling faces) predict rewarding or reinforcing outcomes, the vmPFC integrates affective valuations of specific stimuli made by the amygdala and striatum with inputs from other regions, and the IC is thought to map ascending viscerosensory inputs from the body. Our findings suggest that participants with greater dispositional mindfulness have enhanced emotion processing due to the stronger recruitment of these substrates.

Limitations and Future Research
Our study had a modest sample size for an imaging investigation; thus, the results are preliminary in nature and need to be replicated with a larger sample. Our participants found CALM faces (usually considered a neutral condition and used as a basis for comparison) to be ambiguous. This was evident from the postscan debriefing interviews, in which many participants reported experiencing the neutral expression as fearful or angry. This required excluding the CALM trials from the analysis, effectively eliminating the neutral condition, and contrasting positive and negative emotional face trials directly. In the current study, we only assessed participants at a single timepoint, soon after admission to SUD treatment. Thus, we do not know how emotion regulation ability and underlying neural processes may change with treatment. An additional limitation of this study is that we relied on self-reported stress, depression, and anxiety symptoms, rather than a clinician-administered assessment.
It is important to highlight the specificity of our sample. Given that the study was limited to female participants, it is uncertain whether our findings are generalizable to men. However, given the documented gender disparity in access to SUD treatment Marsh et al., 2021), focusing on women with SUDs can yield insights on women-specific treatment issues. Moreover, compared to men with SUDs, women with SUDs experience significantly higher rates of trauma exposure (Daigre et al., 2015), and traumatic stress has been linked to higher levels of emotion dysregulation (Mandavia et al., 2016). Also, it is unclear if the relationships between dispositional mindfulness and stress, depression, and anxiety would generalize to a population without SUDs. Moreover, additional research is necessary to determine if the negative association between addiction severity and dispositional mindfulness can be extended to suggest a protective quality of mindful disposition against SUD. Finally, although the education level of the study participants was similar to other studies focusing on SUD patients (Ahmadi et al., 2019;Ling et al., 2014), it was lower than in the general population. Prior research found no correlation between education level and task performance or with fMRI activity (Salo et al., 2013). However, it is not clear if the education level affects the relationship between dispositional mindfulness and other constructs we examined.
Our study adds to the emerging literature on the possible protective effects of dispositional mindfulness on key cognitive features, such as emotion regulation, related to SUD treatment outcomes and relapse prevention. Further research is needed to assess the neural effects of mindfulness-based interventions on emotion regulation for SUD population and their potential for improving treatment outcomes such as rates of relapse.
Author Contribution VD: designed and executed the study, performed the data analyses, and wrote the paper. TP: assisted in study execution and data analysis. JM: provided input in study design and data analysis. HA and DB were co-PIs on the parent study from which the data was drawn. All authors collaborated in the writing and editing of the final manuscript and approved the final version of the manuscript for submission.
Funding Funding support was provided by a grant from the National Institute on Drug Abuse (5R01DA038648 to H. Amaro) that was cosponsored by the National Institute on Alcohol Abuse and Alcoholism. T.P. received funding from the NSF GRFP (DGE-1418060). The ideas and opinions expressed herein are those of the authors and endorsement of those opinions by funders is not intended nor inferred.

Data Availability
The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request. All neuroimaging thresholded contrasts reported here are publicly available on NeuroVault (https:// www. neuro vault. org/ colle ctions/ 11918/).

Declarations
Ethics Approval This study was performed in line with the principles of the 1964 Helsinki Declaration and was approved by the University of Southern California Institutional Review Board.
Informed Consent All study participants provided written informed consent.

Conflict of Interest The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.