This item is only available as the following downloads:
THIS IS YOUR BRAIN ON MINDFULNE SS: DISPOSITIONAL MINDFULNESS AND NEURAL ACTIVITY IN ATTENTIONAL NETWORKS BY NATALIE PAUL A Thesis Submitted to the Division of Social Sciences New College of Florida In partial fulfillment of the requirement of the requirements for the degree Bachelor of Arts Under the Sponsorship of Dr. Gordon Bauer Sarasota, Florida May, 2009
ii Acknowledgements I am grateful to Vanderbilt University, Dr. Merida Grant, and Michael Treadway for providing me with access to data, software to analyze the data, and guidance in fMRI theory and analysis. I would also like to ac knowledge Dr. Bauer, for his dedication as an adviser and for challenging me to think critically about ne uroimaging and mindfulness. I also would like to thank Baron Worsha m, for providing unconditional support and assistance navigating the logistics of living in Nashville while conducting my thesis research. I am also grateful for the financia l support provided to me by the New College Council of Academic Affairs, the New College Division of Social Sciences, and the New College Student Research and Travel Program.
iii Table of Contents Acknowledgements ii Table of Contents iii Abstract iv Introduction 1 Methods 42 Results 49 Discussion 52 References 59
iv THIS IS YOUR BRAIN ON MINDFUL NESS: DISPOSITIONAL MINDFULNESS AND NEURAL ACTIVITY IN ATTENTIONAL NETWORKS Natalie Paul New College of Florida, 2009 ABSTRACT Mindfulness is the basis of behavioral th erapies treating mental illness. Evidence suggests mindfulness is attentional skill-set composed of two skill s, concentrative and receptive attention. It was hypothesized concen trative attention scores would correlate with activity in the dorsal network and receptive attention scores would correlate with activity in the dorsal and ventral networ ks. Functional magnetic resonance imaging (fMRI) was used to test this relationship. Pa rticipants took a mind fulness inventory and completed an attentional task in the fMRI s canner. Receptive attention correlated with activity in both networks as hypothesized. Conc entrative attention did not correlate with activity in the dorsal network. These results s upport an attentional conceptualization of mindfulness, although the hypothe sis about mindfulness skills and specific attentional networks was only partially supported. Dr. Gordon B. Bauer Division of Social Sciences
1 Introduction Mindfulness is a set of skills derived hist orically from Buddhist meditation. It is a critical component of new, promising behavioral therapies that treat a range of mental illnesses and psychological stress relate d to physical ailments (Spe ca, Carlson, Goodey, & Angen, 2000; Schapiro, Schwartz, & Bonner, 1998; Baer, 2003; Ramel, Goldin, Carmona, & McQuoid, 2004; Teasdale et al., 2000; Williams, Teasdale, & Se gal, 2000; Williams et al., 2008). These illnesses range from depression to borderline personality disorder, which is a pattern of pervasive instability in the domains of interpersonal rela tionships, affect, and self-identity (American Psychiatric Association [ DSM-IV-TR ], 2000). Despite their success, the psychological mechanisms that account for the efficacy of th ese therapies are not well understood or agreed upon (Baer, 2003; Martin, 2007; Bishop et al., 2004; McKee et al., 2007; Vujanovic et al. 2009; Coffey & Hartman, 2008). Ultimately, the goal of mindfuln ess research is to understand what is driving the efficacy of mindfulness therapies and thus improve the therapies. It is unclear whether attentional or affective self-regulatory skills are the basis of mindfulness (McKee et al., 2007; Vujanovic et al. 2009; Coffey & Hartman, 2008), though there is strong evidence supporting mindfulness as an attention skill (T ang et al., 2007; Slagter et al., 2007; WenkSormaz, 2005; Chan & Wollacott, 2007; Vale ntine & Sweet, 1999; Moore, & Malinowski, 2009). Furthermore, the most accepted definition of mindfulness presents it as an attentional skill (Bishop et al., 2004). The attentional definition pr oposed by Bishop et al. (2004) describes mindfulness as a two-component skill-set. The first component is the self-regulation of attention, which is bringing awareness to current experience and maintaining focu s on a specific object (Bishop et al., 2004). The second component is a unique ori entation to experience to allow all thoughts
2 and experiences to enter awareness without a sp ecific object of focus or agenda (Bishop et al., 2004). Thus, these skills can be summarized as concentrative and receptive attention, respectively. Receptive attention poses a conceptu al challenge to many We stern scientists; it is not a state of having no attenti on, but is an expansion of the focus of attention coupled with rapid deployment and detachment of attention. This classification scheme is the most popular one in the literature currently (Lutz et al., 2008; Jha, Krompinger, & Baime, 2007; Ivanovski & Malhi, 2007; Cahn & Palich, 2006 ; Brown, Ryan, & Creswell, 2007; Valentine & Sweet, 1999; Chan & Wollacott, 2007). Though these skills are distinct, it is possible they develop sequentially, and that receptive at tention develops after the cultivation of concentrative attention (Jha, Krompinger, & Baime, 2007). Indeed, long -term meditators may become proficient in both, regardless of the type of mi ndfulness training they have receiv ed. Even if the two skills are fundamentally intertwined, it is possible to assess the two skills separately. I think it will be the most informative to assess the skills independent ly and see how they are related to a dependent variable measuring an aspect of attention rather than compari ng people given receptive attention training to people given con centrative attention training. With an attentional conceptualization of mi ndfulness, it is necessary to discuss current perspectives on attention. Attention is most commonly defined as the process of concentrating on specific features of the environment, or on certain thoughts or activities (Goldstein, 2005). However, it is more informative to define it as a skill-set rather than a unitary construct because a two-skill definition is able to better explain the results of current neuroimaging studies of selective visual attention. Moreover, the skill-set definition to be proposed maps onto mindfulness skills nearly perfectly. Attention is composed of both goal-driven and stimulusdriven attention (Corbetta & Shulman, 2008). Goal -driven attention biases the processing of
3 stimuli by generating and maintain ing top-down signals using goals and expectations of likely outcomes (Corbetta & Shulman, 2008). Stimul us-driven attention detects salient and behaviorally relevant stimuli in the environmen t, especially when una ttended (Corbetta & Shulman, 2008). Goal-driven attention is mediated by a dorsal frontopa rietal network, and stimulus-driven attention is mediated by both th e dorsal frontoparietal network and a ventral frontoparietal network (Corbetta & Shulman, 2008). The ventral network responds particularly well during a breach of expecta tion (Corbetta & Shulman, 2008). The evidence clearly suggests that both the dorsal and ventral networks mediat e stimulus-driven attent ion in the context of behaviorally relevant stimuli. Ho wever, the relationship between a ttentional networks and salient but irrelevant stimuli is unclear. Most experiments merely present the relevant stimulus in an unexpected location rather than in troducing a novel and task-irrele vant stimulus. Independent of the relationship between neural networks and at tentional skills, a stri king similarity emerges from the definitions mindfulness skills and a ttentional skills. In fact, I hypothesize the concentrative attention maps onto goal-driven attention, and receptive attention maps on to stimulus-driven attention. Indeed, behavioral experiments provide suppor t to the attentional conceptualization of mindfulness. These experiments use different te chniques to teach partic ipants mindfulness, one of which is meditation. Mindfulness can be cultivated in many ways. The most common technique is meditation. In severa l experiments, researchers taught participants how to meditate and then gave them attentional assessment s (Tang et al., 2007; Slagter et al., 2007; WenkSormaz, 2005). Afterwards, the participants dem onstrated a strong improvement in attentional performance as assessed by a variety of standardized and well-accepted measures of attention (Tang et al., 2007; Slagter et al., 2007; Wenk-Sormaz, 2005). One study examining the
4 relationship between dispositional mindfulness as measured by the Kentucky Inventory of Mindfulness Skills and attention found higher mindf ulness scores predicted greater accuracy and speed on two standardized measures (Moore & Malinowski, 2009). Other experiments examined the difference between long-term meditators and controls (Cha n & Wollacott, 2007; Valentine & Sweet, 1999). Furthermore, one experiment demonstr ated that participants had attentional skill improvements that were specific to the type of mindfulness traini ng they had, either concentrative or receptive (Valentine & Sw eet, 1999). One other study, however, found no difference between two groups of participants tr ained extensively in either concentrative or receptive meditation (Chan & Wollacott, 2007). More ev idence is necessary to see if these skills are really distinct and correlate with different attentional processes. Brain imaging research aids the scientific community in understanding the relationship between mindfulness and cognition. Experiments on the neural basis of mindfulness have shown inconsistent patterns of neural activity, and have generally focuse d on participants meditating in a scanner rather than engaging in specific cognitive tasks. Because the benefits of meditation lie in the application of mindfulne ss to daily life, studying meditation is not enough to improve our understanding of mindfulness. The way to test how mindfulness actually works is to give participants a series of tasks in the brain scanner, with each re quiring different skills and seeing where the differences arise between peopl e scoring high and people scoring low on a mindfulness inventory. For instance, it would be useful to study partic ipants performance on tasks assessing different aspects of attention, the self -regulation of emotion, and affect labeling. In this way it would be possible to make sp ecific testable hypothe sis about what cognitive mechanism accounts for the beneficial effects of mindfulness. Therefore, more neuroimaging research on the cognitive mechanisms of mindful ness as a skill in daily life on a cognitive task
5 would be helpful to get a clear picture of th e psychological mechanisms involved. The goal of this study is to examine the relationship between mindfulness, measured as a skill, and neural activity on a selective visual attention task to test the attentional conceptualization of mindfulness. Once researchers determine which cognitive mechanisms make mindfulness therapies effective, it may be possible to modify these ther apies to maximize their e fficacy. The first half of this literature review is on the defin ition and cognitive component s of mindfulness, its relationship to other well-unders tood psychological constructs, wa ys to assess it, behavioral experiments on its relationship with attention, an d the clinical relevance of mindfulness. The second half is on the nature of selective visual attention and the neural correlates of selective visual attention. Finally, I will synthesize thes e topics to create a hypothesis relating specific mindfulness skills to activity in distinct neural networks. Conceptualizing Mindfulness Structured skill-based therapy programs, such as cognitive behavioral therapy (CBT) and mindfulness-based therapies are becoming incr easingly popular because they have been standardized, are easy to teach, and have str ong empirical support for their efficacy. Unlike CBT, mindfulness therapies have not been created fr om new operationalized definitions of desired skills, but have been adapted from a pre-ex isting religious traditi on (Hayes & Shenk, 2004). Thus, the efficacious compone nts and mechanisms are largely unknown and not precisely operationalized. For the purpose of therapeutic intervention it is useful to conceptualize mindfulness as a set of skills, which can be learned through a variety of practices in cluding formal sitting meditation, and which result in a mindful state.
6 This mindful state may become an intrinsic part of ones personality, but it must be consistently maintained. I hypothesize what is important for treatment outcomes is not how mindful a state one can achie ve during formal sitting meditation, but rather how well one integrates this state into daily life as a trait, independently of how the individual learned it. It may be easier to practice mindfulness skills in a particular setting and during a particular time designated for the explicit task of meditation than to consistently apply these skills. Ultimately, the goal of mindfulness interventions is to teach skills which pr oduce long-lasting changes in the individual, including the desire and ability to maintain a continuous mindful state. One study indirectly supports the conclusion that applying mindfulness skills to daily life is more important than achieving a mindful st ate during meditation. It does so by suggesting there is actually little relati onship between the state cultivated in sitting meditation and the measures of mindfulness as a trait (Thompson & Waltz 2007). The researchers first instructed novice participants to meditate and then gave th em a survey to measure how mindful they were in practice, the Toronto Mindf ulness Scale (TMS) (Thompson & Waltz 2007). Afterwards, the researchers gave participants a measure of mindf ulness in daily life, either the Five Factor Mindfulness Questionnaire (FFMQ) or the Mindful Attention Awareness Scale (MAAS) (Thompson & Waltz 2007). The researchers counterb alanced the order of presentation of the meditation session and the assessment of mindf ulness in daily life (Thompson & Waltz 2007). They found either extremely small or non-signifi cant correlations between these two measures (Thompson & Waltz 2007). Defining mindfulness as a cognitive skill ha s many advantages. One advantage of the Bishop et al. cognitive definition is that it aims to distinguish the basic cognitive processes from other related positive psychological outcomes. Many researchers who define mindfulness as a
7 construct using behavioral outcomes come to ve ry different conclusions about the nature of mindfulness. These conclusions can ultimately be unified if mindfulness is considered in terms of cognitive mechanisms. One alternative defini tion, proposed by Leary and Tate (2007) includes diminished self-talk, non-judgment, non-doing, and prosocial philos ophical underpinnings. Diminished self-talk, which is ceas ing the constant internal dialogue an individual has in his or her head, may be a result of the ability to focu s intently on the present. Non-doing, which is the process of acting without the desire for an explic it outcome, may result from the ability to take a detached receptive perspective on ones life. Before creating the Kentucky Inventory of Mi ndfulness Skills (KIMS) with its precise cognitive operational definitions Baer (2003) also proposed several specific outcomes to explain the efficacy of mindfulness which ultimately stem from the basic attentional control mechanisms. Baer suggests mindfulness skills he lp by providing exposure to difficult thoughts or emotions, much in the same way programs designe d to treat phobias or post-traumatic stress do (2003). Mindfulness skills, moreover, may teach self-regulation by helping individuals recognize bodily signals, thoughts, or emotions that may precipitate a maladaptiv e behavior, such as noticing when negative thoughts turn into ru mination (Baer 2003). Emotional self-regulation, however, may be a result of r eceptive attention as well. The relationship between mindf ulness and relaxation is complex. While mindfulness may result in a state of relaxation, th e goal is to accept states which ar e not ultimately compatible with relaxation such as autonomic arousal (Baer 2003 ). Ultimately, these mechanisms, like the ones proposed by Leary and Tate (2007) spring from the same source of at tentional control. Assessment of Mindfulness
8 Even more so than operational definitions, it is necessary to have instruments that quantify mindfulness skills to evaluate mindfulnessbased therapies and investigate the cognitive mechanisms responsible for desirable mental he alth outcomes. One such instrument is the Kentucky Inventory of Mindfulness Skills (K IMS) (Baer, Smith, & Allen, 2004). The KIMS measures the tendency to be mindful in daily life. It is based on the mindfulness skill set created by Marsha Linehan in Dialectical Behavioral Therapy (DBT), a program designed to treat borderline personality disorder (Baer, Smit h, & Allen, 2004). The KIMS operates under the assumption that everyone is to varying degr ees mindful in daily life. Though it may seem problematic to assess mindfulness skills (or any skill ) via self-report, it is important to remember that several disciplines of ps ychology revolve around self-report meas ures and to invalidate selfreport measures is to invalidate fi elds such as personality psychology. The KIMS has four subscales, which include both attentional and affective components. These subscales are Observing, Describing, Ac ting With Awareness, and Accepting Without Judgment (Baer, Smith, & Allen, 2004). The Observing subscale measures the tendency to attend a variety of stimuli, including internal phenomena such as bodily sensati ons, cognitions, and emotions, and external phenomena, such as sounds and smells (Baer Smith, & Allen, 2004). This subscale includes items such as I notice changes in my body, such as whether my breathing slows down or speeds up (Baer, Smith, & Allen, 2004). The Observing s ubscale appears intimatel y tied to receptive attention, and this assump tion is at the core of my experimental design. The Describing subscale measures the tende ncy to covertly apply words to observed internal or external phenomena (Baer, Smit h, & Allen, 2004). Describing, however, is not universally accepted as a valid part of the mindfulness construct because it encourages
9 conceptual processing (Baer, Smith, & Allen, 20 04), which mindfulness does not. I agree with this position. The Acting With Awareness subscale may tap into the concentrative attention skill, because it is engaging fully in ones current acti vity with undivided attention, or focusing of attention on one thing at a time (Baer, Smith, & Allen, 2004). The opposite of this skill is the tendency to exist in automatic pilot (Baer, Smith, & Allen, 2004). Several of the items on the KIMS are reverse-scored to reflect living in a stat e of automatic pilot, such as when I do things, my mind wanders and I am easily dist racted (Baer, Smith, & Allen, 2004). The last of these four subscales is the A ccept Without Judgment subscale. This subscale measures the tendency to be nonjudgmenta l or nonevaluative about present moment experience (Baer, Smith, & Allen, 2004). This s ubscale, however, measures the affective counterpart of the receptive attentional skill. As discussed earlier, this skill may actually be a byproduct of cultivating receptive atte ntion. The KIMS is an interesting instrument because it takes into account the multi-faceted natu re of the mindfulness construct, rather than just generating a single mindfulness score. The KIMS thus allows the attentional components of mindfulness to be analyzed separately. Though the KIMS is a self-report measure of a trait, prelim inary evidence suggests these scores correlate positively with successful tr eatment outcomes of an existing mindfulness therapy, Mindfulness-Based Cognitive Therapy (MBCT). MCBT is different from many cognitive therapies in that it doe s not teach participants to challe nge distorted thinking, emotional regulation skills, or interpersonal skills, but emphasizes mindfulness-practice (Baer, Fischer, & Huss, 2006). In one study ten women with binge-eating disorder particip ated in a ten-session program, and were given the KIMS before and af ter the treatment (Baer, Fischer, & Huss, 2006).
10 After treatment the participants showed an im provement on the Beck De pression Inventory-II, a decrease in objective bi nge-eating events, and a substantial improvement in eating concerns as measured on the Eating Disorder Examination (Baer Fischer, & Huss, 2006). In addition, in the post-treatment analysis, there was a significant positive increase in the scores on the Observation and Nonjudgmental Acceptance s ubscales, which were the only two subscales of the KIMS used (Baer, Fischer, & Huss, 2006). Thus, though the KIMS is a measure of a trait in everyday life, it also is relevant to clinical outcomes. In addition to the KIMS, there are other instruments which measure mindfulness. The TMS, mentioned above, measures how mindful one is in the moment as opposed to on a daily basis (Brown, Ryan, & Creswell 2007). This scale may be informative to assess the value of a particular sitting meditation protocol, but the salutary effect s of mindfulness come from how effectively you can apply the skills to daily life rather than how effectively you can achieve a mindful state during a structured practice. Th e Mindful Attention Aware ness Scale (MAAS) also measures how mindful one is duri ng daily life, but it treats mindf ulness as a unitary construct, which can lead to misleading conclusions (Baer, Smith, & Allen, 2004). For instance, the results of the study discussed above by Thompson a nd Waltz (2007) which found no relationship between the MAAS and the TMS may be due to th e fact the MAAS is a unitary construct when mindfulness is not. Baer (2004) conducted a factor analysis on five independently created mindfulness inventories, and found a c onsistent five-factor structure. Mindfulness in Relationship to Other Psychological Constructs Whenever a new psychological construct emerge s, it is often informative to see how it is related to other more well-under stood constructs and to establ ish convergent validity. The NEO Five Factor Inventory (FFI) is a five factor model of personality, including openness,
11 agreeableness, conscientiousness, neuroticism, and extraversion (Baer, 2004). Baer found that the Observe, or receptive atten tion, subscale correlated positively with openness. Moreover, this subscale correlates positively with several subsca les on an emotional intelligence instrument, the Trait Meta-Mood Scale (TMMS) (Baer, 2004). The subs cales included Attenti on to Feelings and Clarity of Feelings (Baer, 2004) Also, the Observe subscale co rrelated negative with a scale measuring difficulty identifying and describing feelings, the To ronto Alexithemia Scale (TAS) with higher scores meaning more difficulty (B aer, 2004). Baer (2004) also found the Act with Awareness subscale, or concentrative attention, subscale correlated negatively with a measure of dissociation, which is a general lack of awaren ess of ones behaviors and present experiences. These results suggest the KIMS is measur ing the skills it intends to measure. Controversy: Mindfulness and Attention As mentioned earlier, the psychological m echanisms responsible for the efficacy of mindfulness-based therapies are not wellunderstood or agreed upon. Though I am proposing mindfulness is primarily an attentional skill, th ere are other theories about what the underlying mechanism is. In one study, participants were in either a worry group, a mind wandering group, or a concentrative attention breathing gr oup (Arch & Craske, 2006). The worry group was told to worry deliberately, the mind wandering group was told to think about whatever came to mind, and the concentrative attention breathing group followed a breathing exercise adapted from the Mindfulnesss-Based Stress Reduction cour se (Arch & Craske, 2006). All participants given assessments of affect, shown sequences of slides, then assessed again for affect and their ratings of the slides (Arch & Craske, 2006). The slides were negativ e, neutral, or positive (Arch & Craske, 2006). After the displa ys, participants were given th e option of watching a sequence of slides or quitting (Arch & Craske, 2006). The optional sequence was composed of the most
12 negative slides (Arch & Craske, 2006). Researchers found part icipants in the mindfulness group rated neutral slides as more positive and were more willing to watch all the optional negative slides (Arch & Craske, 2006). In addition, the par ticipants in the mindful ness group had a flatter, less varied profile on a measure of negative affect (Arch & Cras ke, 2006). This article suggests that concentrative attention training in the absence of any specific emotional regulation techniques still pr oduces emotional regulatory effects. Research with the KIMS has been studied in the context of negative affect and anxiety sensitivity and has found similar results (McK ee et al, 2007). Anxiety sensitivity is a dispositional variable related to avoidance and fear of anxiety symptoms, and is distinct from trait anxiety (McKee et al, 2007). In one study, participants were given the MAAS to measure mindfulness, the PANAS to measure negative affectivity, and the Anxiety Sensitivity Index (McKee et al, 2007). Higher level of negative affectivity were associated with lower levels of Awareness (concentrative attention) and e xperiential acceptance (McKee et al, 2007). Participants with higher levels of negative affectivity had a lim ited ability to focus and direct attention (McKee et al, 2 007). Higher levels of anxiety sensit ivity were associated with lower levels of concentrative attent ion and experiential acceptance (McKee et al, 2007). These results suggest mindfulness is related to both attentiona l skill and experiential acceptance, an emotional self-regulatory strategy. Post-traumatic stress disorder symptoms present a unique opportunity to study the relationship among emotional self-regulation, atte ntional self-regulation, and mindfulness. PTSD is a failure of self-regulation, and is related to emotional acceptance of events as well as the attentional component of redir ecting attentional focus from a disturbing thought or emotion elsewhere (Vujanovic et al., 2009) One study gave participants the KIMS and then gave them
13 measures of the number of traumatic events, se verity of events, and intensity of symptoms (Vujanovic et al., 2009). The sy mptoms were defined as re-e xperiencing traumatic events, avoidance, and hyperarousal. Nonjudgmental acceptance was negati vely correlated with levels of all PTSD symptoms, even when negative affec tivity and trauma exposure were controlled for (Vujanovic et al., 2009). Concentrative attention was negatively co rrelated with level of reexperiencing traumatic events (V ujanovic et al., 2009). This articl e suggests that acceptance, an emotional self-regulatory skill, and concentr ative attention are both related to PTSD symptomatology, though acceptance may be more important for the treatment of PTSD (Vujanovic et al., 2009). It is al so possible that mindfulness is re lated to different disorders in different ways, and perhaps the attentional component is more relevant for disorders characterized by different cognitive deficits. Another study used a randomized controlled protocol to test both the relationship between meditation experience and emotional regulation as well as the effect of mindfulness treatment on performance on a cognitive task test ing emotional self-regul ation (Ortner et al., 2007). In the first study, particip ants with experience in medita tion were recruited from the community and asked to quantify how long they had been meditating (Ortner et al., 2007). For the task, participants categorized tones presented individually after affective slides, and reaction time was used as a dependent measure (Ortner et al., 2007). Participants were given measures of both state mindfulness (using the TMS) and trai t mindfulness (using the MAAS) (Ortner et al., 2007). Researchers subtracted baseline responding to neutral slides from the affective slides to measure interference (Ortner et al., 2007). Duration of experi ence with meditation and state mindfulness correlated negatively with interfer ence (Ortner et al., 2007). In the second study, participant were given one of the following: seven weeks of mindfulness meditation training,
14 seven weeks of relaxation medita tion training, or no tr aining (Ortner et al., 2007). After training, only participants in the mind fulness meditation group showed a significant decrease in interference (Ortne r et al., 2007). The attentional conceptualization also has s ubstantial behavioral ev idence. Most of the research involves teaching participants med itation, because it is the most common method of cultivating mindfulness. Mindfulness does not exis t in a vacuum; it is th e end result of training. Though there are other methods of cultivating mindfulness, such as using guided CDs (Tang et al., 2007), meditation is the easiest way to teach mindfulness. There is some evidence suggesting the different types of mindfulness training, focusing on either concentrative or receptive attention skills, affect performance differentially on attentional tasks (Val entine & Sweet, 1999). In several experiments, researchers taught participants daily mi ndfulness skills independent of formal sitting meditation practi ce, and other studies focus on sitting meditation practice. The results of the behavioral studies to be discussed below sugge st both forms of training are achieving similar behavioral outcomes. One experiment, broadly using a concentr ative attention techni que, found improvement in a behavioral measure of conf lict monitoring, or the ability to effectively filter out distracters (Tang et al., 2007). Integrative Body-Mind Training (IBMT) is a mindfulness training program that entails following verbal instructions on a CD including guided imagery, posture adjustment, and mindfulness training (Tang et al., 2007). Beca use this technique invo lves focusing attention on verbal directions on a CD, it is concentrative attention mindfulness skill training. This experiment gave undergraduates a behavioral m easure of attentional skills, the attentional network task (ANT), gave them five days of IBMT, and then re-administered the ANT (Tang et al., 2007). The ANT is a task, in which partic ipants must identify the direction a central
15 horizontal arrow is pointing in without being di stracted by two surrounding arrows pointing in either the same or opposite direc tion relative to the central arro w (Tang et al., 2007). Thus, this task involves executive attention, which is the ability to voluntarily filter out distracters. Tasks like the ANT are not very difficult, and thus the most useful measure is reaction time. Participants given the IBMT had si gnificantly faster reaction times than the controls (Tang et al., 2007). These conclusions support my hypothesis that concentrative attention training leads to improvement in executive functi on and goal-direct ed attention. Another experiment examined the relationship between type of meditation, level of meditation expertise, and performance on two atte ntional tasks, the Stroop Task, and the GlobalLocal Letters task (Chan & Woollacott, 2007). This experiment is a corre lational study rather than teaching mindfulness training and then testing attentional changes (Chan & Woollacott, 2007). Participants were either meditators or controls, and meditators were grouped in terms of concentrative or opening up (receptive) style (Chan & Woollacott, 2007) The concentrative style group was composed of Transcendental Me ditation, Sufi Meditation, and Hindu Meditation (Chan & Woollacott, 2007). The receptive style group was compos ed of Vipassana or Tibetan Buddhist Meditation (Chan & Woollacott, 2007). Le vel of meditation experienced was measured by self-report of the number of minutes of meditation per day and the total lifetime number of hours of meditation (Chan & Woollacott, 2007). The Stroop Task entails readi ng color names printed in a diffe rent color ink and name the color of the ink aloud (Chan & Woollacott, 2007). The greater the number of words read, the less Stroop interference and the greater th e ability of the participant to direct his or her attention. The Global-Local letters task is a task in which a glo bal letter is composed of many smaller local letters, and the participants must press a button to indicate either what the global letter is or what
16 the local letter is, dependent upon the condition (C han & Woollacott, 2007). The global and local letters were either congruent (i .e. both the local and global lett ers were the letter s) or incongruent (i.e. the glob al letter was h and the local letter was s). As expected, amount of meditation per day was positively correlated with number of items read on the Stroop task, suggesting me ditation improves executive function. However, there was no difference between the two styles of meditation. On the Gl obal-Local letters task, minutes per day of meditation experience was nega tively correlated with reaction times on all the global conditions (Chan & Wooll acott, 2007). This correlation mean s that the more participants meditated per day, the faster reaction time they had on the global conditi on (Chan & Woollacott, 2007). Once again, researchers found no differen ce between the types of meditation (Chan & Woollacott, 2007). When comparing controls to me ditators, meditators had faster reaction times than controls in all three global conditions (Chan & Woollacott, 2007). Thus, number of minutes of meditation per day predicted a superior attentional performance, as indexed by Stroop interference and reaction times in the Global-local letters task (Chan & Woollacott, 2007). Thes e results support the hypothesis that meditation training, even when independent of a clinical setting, causes improvement on attentional tasks. The Stroop task requires participants to effectively block out distracters, th ereby indexes concen trative attention. The Global-Local letters, by contrast, taps into th e ability to fluidly shift between perspectives, and thus indexes receptive attention. This experiment raises an interesting questi on about the nature of the relationship between concentrative and receptive attention and how functionally independent they are. If the types of meditation are so differe nt, why are participants with l ong-term experience in both types performing so similarly? I hypothe size the reason the researchers didnt find any difference was
17 because individuals extensive e xperience in receptive attention style may require proficiency with concentrative attention (Jha et al., 2007). This experiment might have resolved this problem if they had used an assessment instrument lik e the KIMS. If experienced meditators showed no difference in their scores of concentrative and receptive attenti on respectively, it would only be logical that there would be no difference between groups on behavioral measur es trying to isolate one skill. In addition, one may naturally cultivat e receptive attention with extensive training in a concentrative attention style me ditation. Though long-term training may lead to the development of both skills, short-term training in the context of a therapy may lead to the development of one skill in isolation. In addition, this study di d not really qualify how much time on average participants spent meditating or how much experience they had on average (Chan & Woollacott, 2007). The researchers only provided a range of va lues for each parameter (Chan & Woollacott, 2007). Perhaps with more information to paint a portrait of the average participant in the study, a different explanation would emerge. One experiment demonstrates that receptive attention training may cause a different kind of attentional improvement than the improve ment Chan & Wollacott (2007) found with their concentrative attention training (Sla gter et al., 2007). In this experiment, participants were given an attentional-blink task, received three mont hs of receptive attention meditation (Vipassana) training, and then were given an attentional-bl ink task again (Slagter et al., 2007). This experiment has a very strong design because it has a control group and used a long period of mediation training. The attentional blink refers to the tendency to miss a second target, when two targets are presented embedded in a rapid stream of information (Slagter et al., 2007). This effect only occurs when the second target is presen ted within 500 milliseconds of the first target (Slagter et al., 2007). The theory was that if pa rticipants learn to effectively detach from a
18 stimulus without becoming absorbed in it, they will have a lower rate of missing the second target when presented within the critical time fram e (Slagter et al., 2007). In fact, the practitioner group showed a shorter attentional blink and a greater detection rate of the second target than the control group (Slagter et al., 2007). It is importa nt to note that this st yle entails cultivating concentrative attention in the early phases of the training. One way of explaining the results of these attentional studies is that improvements in attention may be due to reductions in arousal. If controlling for a particip ants level of arousal eliminates changes between a control and mindful ness-training group on an attentional measure, then mindfulness is not really an attentional skill. If arousal mediates the relationship between mindfulness and attention, the changes due to mindfulness are actually related to selfregulation and a reduction in arous al. One study tested this possibi lity explicitly (Wenk-Sormaz, 2005). It examined the relationship between a concentrative style of meditation (Zen), performance on the Stroop task, and arousal (Wenk-Sormaz, 2005). In this experiment, participants were assigned to one of three conditions: medita tion, learning, or rest (WenkSormaz, 2005). In the learning cond ition, participants had to use memory techniques to learn the presidents of Yale university, specifically the method of loci (Wenk-Sormaz, 2005). Participants created a mental visual image of a place in whic h they have lived and represented each of the previous presidents in a specific region (Wenk-So rmaz, 2005). In the rest condition, participants, were told to simply rest for 20 minutes (Wenk-Sormaz, 2005). The experimental group was given Zen meditation training (Wenk-Sormaz, 2005). Wenk-Sormaz (2005) used a test-retest protocol. As well as having three groups, the ex perimenters also measured the galvanic skin response, a measure of arousal, during the first an d last three minutes of the exercise (meditation,
19 learning, and rest) (Wenk-Sormaz, 2005). They ca lculated the measure of arousal by taking the average of these two points during the exercise (Wenk-Sormaz, 2005) The participants in the meditation group had significantly less Str oop interference than participants in the other groups, and changes in arousal did not change the relationship (WenkSormaz, 2005). In this context St roop interference was defined as the difference in reaction time for an incongruent word-ink pairi ng and a baseline reaction time, in which the participant must complete a block of trials composed of multip le xs in one color (Wenk-Sormaz, 2005). Thus, this experiment demonstrates another case in whic h participants were trained in a concentrative attention style of meditation, given a concentrativ e attention task, and performed better than two control groups Furthermore, a study actually did find differen tial results for particip ants as a function of meditation type. This experiment used the Wilkin s Counting Test, in which participants were presented with a repetitive sequence of tones a nd were asked to count them (Valentine & Sweet, 1999). When the frequency of tone presentation is fairly low, it is easy to develop expectations of when the next tone will come (Valentine & Sweet, 1999). Errors on the slow frequency portion are a result of losing attention out of bor edom or distraction (Valentine & Sweet, 1999). However, when the frequency of tone presenta tion is fairly high, counting the tones becomes more difficult and requires participants to be ab le to detach their attention rapidly from each tone, and each tone is difficult to predict (Valentine & Sweet, 1999). The experiment looked at experienced med itators, of both recep tive and concentrative styles, and controls (Valen tine & Sweet, 1999). In addition, the meditators were given questionnaires to index their meditation experien ce, as 24 months or less, or greater than 25 months (Valentine & Sweet, 1999). The researchers predicted both groups of meditators would
20 be more accurate than non-meditators, and when the frequency of the tones was high, the receptive attention meditators would be more accurate (Valentine & Sweet, 1999). The results confirmed this hypothesis. For the low frequency task, meditators were more accurate than controls, those with greater experience were more accurate than those with less experience (though this was not broken down by type), and both groups of meditato rs were equally accura te (Valentine & Sweet, 1999). The equal accuracy result is in line wi th result of Chan & Woollacott (2007). If participants must develop con centrative before receptive atten tion, the Stroop task would not show a difference between types. In addition, the Global-Local letters may not have been sensitive enough to detect differences between types. For the high-frequency task, receptive attention meditators were more accurate than concentrative attention meditators (Valentine & Sweet, 1999). This study, however, did not teach pa rticipants meditation, but used participants already having this training. One study, however, did not find differences in attentional performance between a control and a mindfulness experimental group (Ande rson et al., 2007). Participants took an eightweek MBSR while participants in the control group did not get a ny sort of training (Anderson et al., 2007). This experiment has a very strong design, because the MBSR course is relatively long, they had a control group, they used a wide array of attentional measures. Both groups were given standardized measures of atten tion, including a measure of sustaine d attention, the ab ility to shift attention flexibly between stimuli, an object dete ction task, and the Stroop test (Anderson et al., 2007). Presumably, the sustained attention task and Stroop test were intended to measure concentrative attention and the te sts of attention shifting and obj ect detection were intended to tap into receptive attention. The mindfulness group did not demonstrate improvements in
21 performance relative to the c ontrol group after the course (Anderson et al., 2007). One explanation of these results is th e nature of the MBSR course. It involves elements of cognitive therapy, yoga, and relaxation (Cha mbers, Yee Lo, & Allen, 2007). A study using an intensive pure meditation program without the othe r added elements of MBSR did find improvements between a mindful ness group and a control group. In this study, participants in the mindfulness group went on a ten day mediation retreat, during which they meditated for 110 hours (Chambers, Yee Lo, & A llen, 2007). Participants were given a measure of working memory, the Digit-Span Backwards test and an internal switch ing task. The internal switching task measured participants ability to keep a mental tally of words belonging to one of two categories while the words were presented in random order (Chambers, Yee Lo, & Allen, 2007). The researchers found that participants in the meditation group had significantly faster reaction times after the retreat relative to controls for the internal switching task (Chambers, Yee Lo, & Allen, 2007). They also demonstrated im proved working memory (Chambers, Yee Lo, & Allen, 2007). A different study examined the relationship between dispositional mindfulness, like the current study does, and performance on an atte ntional measure. Participants were either experienced meditators or meditation nav e controls (Moore & Malinowski, 2009). The meditators were either in intermediate med itation classes or recently completed a six-week introductory course (Moore & Malinowski, 2009) All participants completed both the Stroop task and the d2-concentrati on and endurance task (Moore & Malinowski, 2009). This task requires participants to scan a sheet of paper for targets, mark through them, and ignore similar distracters (Moore & Malinowski, 2 009). The target was either the letter d or the letter b, and had a unique configuration of quotation marks around it (Moore & Malinowski, 2009). All
22 other targets had to be ignored. Participants were given the KIMS to measure dispositional mindfulness (Moore & Malinowski, 2009). Total mindfulness scores correlated positively with the total number of words read on the Stroop task, and the number of correct targets identified on the d2-concentration and endurance task when the data from both groups were plotted together (Moore & Malinowski, 2009). In addition, the meditators had significantly higher mindfulness scores the controls (Moore & Malinowski, 2009). This study implies measures of dispositional mindfulness can distinguish meditators from c ontrols, and that these measures do correlate positively with improved attentional control (Moore & Malinowski, 2009). Thus, the research strongly suggests mi ndfulness training causes improvements in attentional performance, and KIMS scores correlate positively with superior attentional performance. The relationship between receptive a nd concentrative attention is less clear, though there is support for the notion th at receptive attention only develops after c oncentrative attention. The training methods to develop receptive a ttention inevitably start with exercises in concentrative attention. The KIMS presents an opportunity to study both components of mindfulness independently, and see if there is a correlation between the two skills, using the two subscales. Clinical Relevance of Mindfulness Mindfulness is clinically relevant because it is the basis of se veral new behavioral therapies that are in the process of being investigated. MindfulnessBased Stress Reduction (MBSR) is probably one of the most well-know n mindfulness programs (Ivanovski & Malhi, 2007; Baer 2003). It is an eight to ten week program, composed of intense two to two and a half hour weekly meetings, and one fu ll day retreat (Ivanovski & Malhi, 2007). Rather than trying to change the individuals cognitions or teach the participant to cognitively confront distorted
23 cognitions, MBSR changes the participants relationship with his or her thoughts (Baer 2003; Baer 2006). The participant learns to see thoughts as fleeti ng mental events and becomes comfortable experiencing them, leading to a re duction in stress and improved emotional selfregulation (Ivanovski & Malhi, 2007 ). MBSR was developed to tr eat chronic pain and stress disorders; participants label their feelings wi th words rather than avoiding them (Baer, 2003). Though there is an abundance of research on mindf ulness-based therapies, much of the research is not adequately controlled. Research will only be considered here if it has a control group. Control groups here are defined as an activ e treatment group (either placebo or empirically demonstrated), a wait-listed control group, or an active but non-speci fic and non-controlled treatment (called treatment as usual) One experiment using random assignment examin ed levels of mood disturbance in cancer patients (Speca, Carlson, Goodey, & Angen, 2000) after an MBSR course. They found a significant improvement in depressive symptoms, em otional irritability, an ger, and symptoms of stress compared to a wait-listed control group (Speca, Carlson, Goodey, & Angen, 2000). An analysis of this study by Baer (2003) f ound a substantial effect of treatment ( d = 0.60). In addition, a study examining stress levels in me dical students and prem edical students using random assignment found a significant improvement on measures of depression, state and trait anxiety, empathy in comparison to a wait-listed cont rol group (Schapiro, Schwartz, & Bonner, 1998). Baer (2003) calculated a me dium effect for this study ( d= 0.50). A third study examining symptoms of depression, anxiety, and rumination in a group of vetera ns with a history of lifetime depression found a significant decrease in rumina tion after the course in comparison to waitlisted controls (Ramel, Goldin, Carmona, & McQ uoid, 2004). The researchers calculated a very large effect size ( d = 1.47) (Ramel, Goldin, Carmona & McQuoid, 2004). This research
24 demonstrates that MBSR has the potential to be a very successful treatment, and that further research is necessary. As of this review, two studi es with an active standardiz ed treatment control group was found examining the efficacy of MBSR. This st udy used a population of individuals with moderate to severe psoriasis and used rate of skin clearing as a depende nt measure, as assessed by a blind evaluator (Kabat-Zinn et al., 1998) An MBSR plus light treatment group was compared to a light treatment alone group (Kabat-Zinn et al., 1998). Researchers found participants in the MBSR condition had a significantly faster clearing rate th an participants in the light treatment alone group (Kabat-Zinn et al., 1998). The second study compared MBSR to CBSR, which is a program based on CBT that is aimed at changing and replacing distorted cognitions (Smith et al., 2008). This study measured depressi on (using the BDI), psychological well-being, perceived stress, percei ved pain level, energy level, and neuroticism (Smith et al., 2008). Within the MBSR group, there was a posi tive increase in well-being and energy and a decrease in depression, neuroticism, binge ea ting, and pain (Smith et al., 2008). Within the CBSR group, there was an increase in well-being and a decrease in perceived stress and pain (Smith et al., 2008). The MBSR group had larger e ffect sizes in the posi tive direction than the CBSR group (Smith et al., 2008). Furthermore, a mu ltivariate analysis demonstrated that there was a greater change on all vari ables in the MBSR group compared to the CBSR group (Smith et al., 2008). The only limitation to this study is that it was not random assignment; community member had short descriptions of the programs and chose which group to be part of (Smith et al., 2008). Mindfulness-based cognitive therapy (MBCT) is another mi ndfulness based intervention, and is specifically an adaptation of MBSR for depression (Bae r, 2003). It teaches individuals
25 with depression to learn to obser ve their thoughts and emotions, in order to try and prevent them from entering the downward cycle of negative affect and cognitive rumination (Baer, 2003). In one randomized study, participants with a history of lifetime depression who were currently in remission but off medication were significantly less likely to experi ence recurrence of depression within the year following completion of MBCT th an controls who were instructed to continue Treatment As Usual (TAU) (Teasdale et al., 2000). Interestingly, for patients with three or more episodes of depression, MBCT reduced relapse ra tes, but not for patient s with two or fewer episodes (Teasdale et al., 2000). Baer (2003) calculated a medium-large effect size ( d = 0.60) for this study. In another randomized study examin ing participants with recently remitted depression who were not on medication, participan ts who received MBSR had greater recall of specific autobiographical memories than participants in the TAU group (Williams, Teasdale, & Segal, 2000). What is interesting about this study is that usually individuals with a history of depression have impaired recall of specific autobiographical memories whether they are currently experiencing depression or not (W illiams, Teasdale, & Segal, 2000). Baer (2003) calculated a strong effect of the mindfulness treatment (d = 0.70). In another randomized trial using both bipolar and unipolar depression patients currently in remission, researchers found decreased depression in the MBCT group relative to wait-listed controls (Williams et al., 2008). Thus, MBCT is a promising treatment wh ich should continue to be studied. One study using an active treatment group studying MBCT was found (Teasdale et al., 2008). A randomized controlled study tested MBCT while participants were tapering off medication against maintenance antidepressant medication (m-ATM) (Teasdale et al., 2008). It used time to relapse, residual symptoms, and quality of life as dependent measures (Teasdale et al., 2008). This study is the first MBCT study using an active treatment arm, and has several very
26 strong methodological features (Teasdale et al., 2008). In addition, the raters on the dependent measure were blind to group assignment, and two raters coded each taped interview (Teasdale et al., 2008). A trained therapist watched all group ther apy sessions to ensure protocols were being adequately followed (Teasdale et al., 2008). In the MBCT arm, the relapse rate was 47% while in the ATM arm the relapse rate was 60%. MBCT significantly reduced residual symptoms as measured on the clinician rated Hamilton Rati ng Scale for Depression (HRSD) and the selfreport Beck Depression Inventor y (BDI) (Teasdale et al., 2008) In the MBCT group, 75% of participants successfully tapered off medication (Teasdale et al., 2008). Another common therapy is Acceptance and Commitment Therapy (ACT). ACT does not try to change the content of thought either. It uses mindful acceptance to target experiential avoidance and thought suppressi on (Hayes, 2004). One study examined the effect of ACT on frequency and duration of epilept ic seizures and quality of life (Lundgreen, Dahl, Melin, & Kies, B, 2006). The experimental group received ACT in addition to behavioral seizure control technology, and the control group re ceived supportive therapy (ST) (Lundgreen, Dahl, Melin, & Kies, B, 2006). Supportive therapy is a placebo treatment in which th e therapist asks participants to reflect on their experiences and feelings regarding having seizures providing a supportive environment but providing no advice (Lundgreen, Dahl, Melin, & Kies, B, 2006). Participants in the ACT group reported fewer and shorter seizures than partic ipants in the control group (Lundgreen, Dahl, Melin, & Kies, B, 2006). They also reported grea ter life satisfaction (Lundgreen, Dahl, Melin, & Kies, B, 2006). A second study examined the relationship between ACT and workplace-related stress, using the general health questi onnaire (GHQ), a measure of gene ral mental health, and Beck Depression Inventory (Bond & Bunce, 2000). There were three groups: ACT, an inactive control
27 group, and an Innovation Promotion Program (IPP) (Bond & Bunce, 2000). The IPP taught participants to seek new creative solutions to problems. Both groups showed improvements on the BDI and GHQ relative to controls (Bond & Bunce, 2000). The ACT group showed a significant improvement over the IPP tr aining on the GHQ (Bond & Bunce, 2000). A different of a non-clinical population f ound an improvement in broad psychological health relative in an ACT group relative to cont rols in a CBT control group (Lappalainen, 2007). Both groups were given the BDI and a genera l checklist measuring psychopathological symptoms (SCL-90) before and after the inte rvention (Lappalainen, 2007). There was a medium between-group effect size after treatment a nd at follow-up favoring ACT relative to CBT (Lappalainen, 2007). In addition, there was a medium effect size again favoring ACT for the BDI scores (Lappalainen, 2007). Therefore, mindfulness ther apies show promise, though a dditional resarch is needed. Unfortunately, the vast majority of mindfulnessbased therapy studies us e uncontrolled designs. Recommendations for improvement can be found in the discussion section of this literature review. Cognitive Components of Selective Visual Attention Given the studies on the eff ect of mindfulness training on at tention, and the efficacy of mindfulness interventions for depression, it is li kely that there is an attentional mechanism driving the efficacy of these interventions. To examine the neural correlates of mindfulness, however, it is necessary to understand the neural correlates of selectiv e visual attention. As described earlier, attention is a cognitive proce ss that is composed of stimulus-driven and goaldriven attention. (Corbetta & Shulman 2008).
28 One example of behaviorally relevant goal-dr iven attention is no ticing a stranger in a green hat when you are looking for a friend in a green shirt in a crowd; it is based upon expectation (Corbetta & Shulman 20 08). The breach of expectation refers to the fact it shares similar properties with the target stimulus and is occurring in an unexpected location. On the other hand, irrelevant stimulus-driv en attention is a function of stimuli salience, and may occur in response to loud noises, motion, color contrast or changes in luminance (Corbetta & Shulman 2008). It is controversial whether task-irrelevant but salient distra cters or targets activate the same network as distracters or targets which are relevant but not salient (Corbetta & Shulman, 2008). Properties of Selective Visual At tention: Biased Competition Model Apart from its cognitive components, the proper ties of attention are a direct result of its limited capacity. Because attention is limited, st imuli compete for neural representation and processing, and the organism must have the ability to filter out irrelevant stimuli (Desimone & Duncan 1995). There is abundant behavioral evidence that suggests when multiple stimuli are presented together, accuracy and speed diminish (Desimone & Duncan 1995). Indeed, accuracy is restored when stimuli are presented in se quence rather than simu ltaneously (Desimone & Duncan 1995). Neural evidence from single-cell recording in monkeys also supports the conclusion that attentional resources are limited and stimuli mu st compete for them. When two objects are presented within a center-surround neurons receptive field, one effective stimulus and one ineffective stimulus, the activity of the neuron is suppressed (P essoa, Kastner, & Ungerleider 2003). An effective stimulus here refers to one that evokes the maxima l firing of the neuron, such as with contours in V1. In addition, as information travels from receptors on the retina to
29 the cortex, the size of the receptive field incr eases from 0.2 degrees to 25 degrees. Thus, inessential information is lost at each st age of processing (Desimone & Duncan 1995) and receptive fields are the basic unit of competition. Properties of Selective Visual Atten tion: Top-Down and Bottom-Up Mechanisms The two cognitive mechanisms which bias the competition for object processing are the top-down and the bottom-up mechanisms. The to p-down mechanism biases competition when stimuli in the environment are relevant to th e task at hand (Desimone & Duncan 1995). The topdown mechanism is analogous to con centrative attention. In fact, dire cting attention to a stimulus can modify the neural response to the stimulus. When monkeys are given a spatially directed cue to the location of the target, the neuron responds as if the receptive field effectively shrunk around the target during single-ce ll recording studies (Pessoa, Kastner, & Ungerleider 2003). The single-cell studies with monkeys are the most direct way to measur e neural activity. Blood oxygen level dependent (BOLD) signals are an indirect method for measuring neural activity; it assumes greater activity reflects summed activity of many local neurons. However, these signals measure inputs to an area as well as local cort ical processing, includi ng both inhibitory and excitatory interneurons (Pessoa, Kastner, & Ungerleider 2003). The bottom-up mechanism biases competition when they contain salient features, like local discontinuities, novelty, movement, colo r contrast, etc (Desimone & Duncan 1995). Topdown and bottom-up processes are directly related to goal-driven attention and stimulus-driven attention. Goal-driven attention is probably only influenced by top-down processes, and stimulus-driven attention is probably influenced by both processes. Stim ulus-driven attention may be influenced by top-down processes if the st imulus is relevant or to the task at hand.
30 The relationship between atte ntion and visual processing, ho wever, is not simple. Some unattended stimuli do undergo processing, and thus pr ocessing is not limited to attended stimuli alone. The degree to which unattended stimuli are processed depends on the available attentional load. Even so-called pre-attentive tasks, such as tasks requiring a pa rticipant to recognize a unique stimulus within an arra y of different but homogenous di stracters, require attention (Pessoa, Kastner, & Ungerleider 2003). Even emotional stimuli, which researchers thought were processed automatically, may be contingent upon available attentional resources (Pessoa, Kastner, & Ungerleider 2002). In one experiment, experimenters used two tasks to isol ate neural responses in the context of emotion while participants were in one of the followi ng two conditions, an attended condition or an unattended condition. In the attended condition, pa rticipants viewed fa ces with different emotional valence and responded with a button-push to indicate the sex of th e central face. In the unattended condition, participants pushed a button to indicate whet her two bars adjacent to the same series of faces were in the same orientatio n and were told to ignore the faces. Researchers found differential activity in response to the fear ful faces in the unattended compared to the attended condition. It was only in the atte nded condition the researchers found amygdala activation. Negative faces are more likely to invo luntarily attract atten tion than neutral or positive faces (Pessoa, Kastner, & Ungerleider 2002). Neural Correlates of Attention The relationship between attention and neuroanatomy is complex and controversial. One of the most widely accepted models postulated tw o types of attention, goal-driven attention and stimulus-driven attention (Corbetta & Shul man, 2008). Goal-driven a ttention biases the processing of stimuli by generating and main taining top-down signals using goals and
31 expectations of likely outcomes (Corbetta & Shulman, 2008). In th is model, a bilateral dorsal frontoparietal network is responsible for goal-driven attention. It is composed of the intraparietal sulcus (IPS), frontal eye field (F EF), and the superior parietal lobe (SPL) (Corbetta & Shulman, 2008). Stimulus-driven attention de tects behaviorally relevant stimuli, especially in an unexpected location (Corbetta & Shulman, 2008). S timulus-driven attention, in the case of behaviorally relevant stimuli, is mediated by the dorsal network as we ll as a primarily rightlateralized ventral frontoparietal network. The ventral network is composed of the temporoparietal junction (TPJ), in ferior frontal gyrus (IFG), and an terior insula (aI) (Corbetta & Shulman, 2008). As mentioned earlier, the relatio nship between salient irrelevant stimuli and attentional networks is unclear. One study sugge sts there is no relationship between salient irrelevant stimuli and the attentional networ ks as defined above (Indova & Macaluso, 2007). Another study suggests it may be mediated by the dorsal network exclusivel y (de Fockert et al., 2004). The dorsal and ventral frontopari etal networks may respectively have intrinsic functional connectivity (Fox et al., 2006a). Indeed, one way to measure functional conne ctivity of networks is to examine spontaneous neural activity that is not task-depe ndent (Fox et al., 2006a). In the brain there are significant neural fluctuations that are unrelated to the stimulus or task at hand (Fox et al., 2006b). These fluctuations, however, are not random but are consistent within a system (Fox et al., 2006b) .Because most of the brains energy is used to maintain internal functions, spontaneous activity may illustrate fundamental underlying patterns of connectivity (Fox et al., 2006b). One experiment found that correlation maps w ithin the ventral network (between the TPJ and ventral frontal cortex) were significantly different from th e correlation maps within the
32 dorsal network (the IPS and FEF) (Fox et al., 20 06a). In addition, the correlations between the two regions in both networks respectively were stronger than correlati ons between any other region pair (Fox et al., 2006a). The results of th is functional connectivity analysis provide an informative and empirically-based map of the dorsal and ventral networks: The areas in red are the ventra l network, and the areas in bl ue are the dorsal network. The topmost left and right images are showing the brain as if it were butterf lied, with the top left image showing the left hemisphere and the top ri ght image showing the right hemisphere. On the top right image, the frontal cortex is pointing to the right. On the top left image, the frontal cortex is pointing to the left. These images show how right-lateralized the venteral network is. In particular, in the top right image, the ventral network is primarily composed of the ventral frontal cortex, including the inferior fr ontal gyrus, and the temporopariet al junction, which is the red region posterior to the central sulcus. The dorsal network is best viewed from the overhead image, which is the top central image. The regi on anterior to the central sulcus contains the
33 frontal eye field, and the region pos terior to the central sulcus contains the intraparietal sulcus and superior parietal lobe. Though these regions may have intrinsic func tional connectivity, they have overlapping functions and do not separate as neatly as Co rbetta and Shulman (2002) originally suggested (Corbetta & Shulman, 2008). One of the most common experimental designs is the classic spatial cueing Posner design. It usually entails an expectation period, followed by a cue, and then a target (Kincade et al. 2005). Du ring the expectation period, no targ et or cue is presented. It is designed to capture neural activ ity during goal-driven attention. In some experiments, the cue period is not preceded by an expectation period and the cue alone is thus intended to capture goal-driven attention. The cue is usually an arrow pointing one way or an asterisk, and the target is then presented on the same hemifield (a c ongruent condition) or th e opposite hemifield (an incongruent condition) (Kincad e et al. 2005; Lepsien & Pollman, 2002; Indova & Macaluso, 2007). This way it is possible examine activity as a result of expectati ons being validated or invalidated (Kincade et al. 2005; Lepsien & Pollman, 2002; Indova & Macaluso, 2007). The participant typically has to respond behaviorally to a featur e of the target, for instance its orientation, with a button (Kin cade et al. 2005; Lepsien & Pollman, 2002; Indova & Macaluso, 2007; Hopfinger, Buonocore,& Mangun, 2000 ). In many of these designs the cue correctly predicts the location of the cue fairly consistently, or about 75-80% of the time (Kincade et al. 2005; Lepsien & Pollman, 2002; Indova & Macaluso, 2007). In addition, some studies have a very salient but irrelevant cue or feature of the cue, to try a nd capture stimulus-driven atten tion (Kincade et al., 2005; Indova & Macaluso, 2007; Hopfinger, Buonocore,& Mangun, 2000; de Fockert, Rees, Frith, & Lavie, 2004).
34 To separate the different patterns of neur al activity associated with attention, many studies use event-related functional magnetic re sonance imaging (fMRI) design. With this kind of design it is possible to separate cue activity from target ac tivity (Kincade et al., 2005; Corbetta 2000; Corbetta 2002). In addition to discussing common designs, it is necessary to understand common methods of analysis. One of the most common ways to an alyze data is through the use of a contrast. A contrast is a method in which activity in one c ondition is essentially st atistically subtracted from activity in another. A simple example woul d be to look at a cue condition and a baseline condition. In the contrast cue baseline, one is subtracting baseline activity (focusing on a fixation point for example) from neural activity while the participant is viewing a cue. This notation is usually written in the literature as cue > baseline rather than cue baseline. The whole-brain analysis is conducted this way for each volumetric pixel (voxel). There are 100,000 voxels in the brain, and each one represents a space of 1 mm3. The articles presented below showcase both typi cal Posner designs and variations on it to illustrate the nuances of the relationship between the dorsal and ventral networks. These atypical experiments go outside the realm of the usual method, for instance by using different modalities (Downar, Crawley, Mikulis, Davis, 2001), incorporating a visual search element without spatial expectations (de Fockert, Rees, Frith, & Lavie, 2004) or creating spatia l expectations without the need for reorienting (Hopfinger, Buonoc ore,& Mangun, 2000). I hypothe size the function of the dorsal network is for both ta sk-relevant stimulus-driven atte ntion and goal-driven attention. In addition, I hypothesize the ventral network is primarily active during stimulus-driven attention, especially when it is relevant. I will first analyze the typical Posner method experiments and propose that seemingly contradict ory results are due to whether studies image
35 targets and cues together (Peelen, Heslenfe ld & Theeuwes, 2004; Indova & Macaluso, 2007)or image them separately (Kincade et al., 2005; Co rbetta et al. 2000; Co rbetta et al. 2002). Afterwards, I will describe the modified Posner designs. One event-related classic Posner design f ound differential activation for the networks when data were acquired for both cue and targ et periods, though they still shared overlap (Kincade et al., 2005). In this experiment there was a cue period, an inte r-trial interval, and a target (Kincade et al., 2005). The cue array wa s composed of a central diamond flanked by eight colored boxes of different colors (Kincade et al., 2005). In the e ndogenous (goal-driven) condition, one half of the diamond changed luminan ce to indicate the likely location of the target (Kincade et al., 2005). This experiment also had an exogenous (stimulus-driven) attention component. In the exogenous condition, seven of th e eight boxes were the same color, with an eighth one a different color (Kinca de et al., 2005). The theory was that the color contrast would involuntarily capture attention and would direct it to that location (Kincade et al., 2005) In the exogenous condition the cue did not predict target location at all. In addition, there was a neutral condition with eight different colored boxes, and no specific cue (Kincade et al., 2005) During the cue period, there was both dorsal and ventral activity. However, there was temporal separation of the networks suggesting different, though related functions. Immediately after cue presentation, the resear chers found dorsal activity (FEF and IPS) (Kincade et al., 2005). This results supports the conclusion that the do rsal network is respons ible for goal-driven attention. Two seconds after this initial activity the researcher s found ventral activity (TPJ and IFG) (Kincade et al., 2005). Given how few st udies use event-related designs and collect imaging data during cue presentation, it is possible the activity that is observed in the ventral network actually comes slightly before target presentation but after cue presen tation. Given the
36 way these two pairs of regions appeared together, it supports the idea that these two networks are distinct, though they share overl apping functions. It is possible there is a gradual transition between usage of the networks between cue presentation and target presentation. When the experimenters analyzed the da ta more thoroughly, some regions showed greater responses to invalid targets than valid targets in the endogenous condition after target presentation. These regions were the FEF (dorsa l) and the TPJ (ventral) (Kincade et al., 2005). These results support the conclusion that both dors al and ventral networks are responsible for stimulus-driven attention, especially invol ving a breach of exp ectation of location. Thus, Kincade et al. (2005) implies the dorsal network is responsible for goal-driven attention and also responds during a breach of e xpectation when it is ta sk-relevant. In addition, the article suggests the ventral network is also responsive to be haviorally relevant violations of expectations in the context of relevance. This study has a few very good methodological features which not all experiments share. It has a fixed inter-stimulus interval betw een the cue and target. A fixed inter-stimulus interval between cue and target al.lows participants to maintain a goaldriven expectation, which is an implicit feature of the dorsal netw ork. The article also monitored eye fixation, supporting the conclusion that FEF activity was not due to saccade activity (Kincade et al., 2005). FEF activity is strongly relate d to saccades, so it is vital to monitor eye fixation. In addition, exogenous and neutral trials were mixed toge ther, and kept separate from endogenous trials, which may have allowed partic ipants to develop goal-driven expectation between blocks (Kincade et al., 2005). Several other experiments which use eventrelated fMRI to separate cue and target activity have come to similar conclusions, reserving a special place for the dorsal network in goal-driven attention, while the st imulus-driven attention requires both networks (Corbetta et al.
37 2000; Corbetta et al. 2002). Corb etta et al. (2000) found IPS activ ity alone during a cuing period, and both TPJ and IPS activity during target presen tation. Corbetta et al. (2002) found cue-related activity in the IPS, and near th e FEF. In addition, Corbetta et al. (2002) both dorsal and ventral activity for target related activity for the invalid > valid contrast. Another basic spatial cueing Po sner experiment examined the effect of stimulus salience on reorienting using a similar design and also cam e to similar conclusions about the relationship between the ventral network and the stimulus -driven attention. This experiment, however, supports the conclusion that the ventral network is preferentially active for stimuli which are relevant and non-salient rather than irrelevant and salient. In this experiment participants were given a cue identifying location of the target, and given the color of the target in advance (Indova & Macaluso, 2007). The cue could also be valid or invalid predicting location; they had to respond to the orientation of the target to test endogenous attention. In addition, sometimes they unpredictably saw a checkerboard at the cued (valid) or uncued (invalid) region, which did not require a behavioral response (Indova & Macalus o, 2007). For the contrast testing invalid target > valid target activity, research ers found activity in the IPS, IFG, insula, and FEF (Indova & Macaluso, 2007). These results support the conc lusion both dorsal and ventral activity are present for stimulus-driven attention, especi ally when it is at an unexpected location. This pattern of response suggests behaviorally relevant shifts of attention require both networks. Interestingly enough, creati ng a contrast with [invalid targ ets > valid targets] [invalid checkerboard > valid checkerboard ] did not reveal any new regions of activity. Though this contrast looks complicated, it is simple in prin ciple. The two contrasts invalid targets > valid targets and invalid checkerboard > valid checkerboard are created to isolate activity related to presentation of a target in an unexpected location that is different from targets in an expected
38 location. The final operation is subt racting the two contrasts, to create a new contrast. The new contrast isolates activity related to the presenta tion of a low-salience but high relevant target from activity related to the pres entation of a high-salience irrelevant target. These results suggest activity resulting from salient irrelevant distracters does not even affect the same attentional networks as behaviorally relevant stimuli (Indova & Maca luso, 2007). Valid, invalid, and checkerboard trials were randomly intermixed. This experiment found no distinction between networks, probably because it did not imag e cue and target periods separately. A third familiar Posner method experiment suggests this same conclusion. This experiment used a familiar Posner design with a spatial cue either predicting target location or not predicting location and cues that were ei ther valid or invalid (Peelen, Heslenfeld & Theeuwes, 2004). In a goal-driven attention condi tion, the cue predicted ta rget location (Peelen, Heslenfeld & Theeuwes, J. 2004). In the stimulus -driven attention condition, the target did not predict location (Peelen, Heslenfeld & Theeuwe s, 2004). This experiment separated cue and target activity, and is also mixed the stimulus -driven and goal-driven trials together so participants could not develop goal-driven expectations (Peele n, Heslenfeld & Theeuwes, 2004). In an analysis examining both stimulus-drive n and goal-driven condition, experimenters found both dorsal (FEF) and ventral activity (IFG, TPJ) (Peelen, Heslenfeld & Theeuwes, 2004) during cue presentation, regardless of whether it was pred ictive or not. Due to the method of analysis, combining both predictive and non-predictive cues, it is unsurprising both networks were active. Interestingly, one study imaging both cue a nd target activity toge ther found ventral activity alone in the i nvalid > valid contrast. This experi ment, on the other hand, used both an arrow and a geometric shape to as a cue. Ob ject-based reorienting may have a different mechanism (Arrington et al., 2000).
39 In contrast with the above typical Posn er methods with valid/invalid conditions and stimulus-driven/goal-driven conditions, other ex periments systematically vary one of these components or collapse one level of variable. One such experiment used a modified Posner paradigm (Hopfinger, Buonocore,& Mangun, 2000) and was designed to primarily test goaldriven attention. Fitting in with previous research, it found prim arily dorsal activity (Hopfinger, Buonocore,& Mangun, 2000). In this experiment par ticipants were cued to determine whether the reversing checkerboard presented at that lo cation contained some grey checks or only black and white checks (Hopfinger, Buonocore,& Man gun, 2000). This design, however, differed from the majority of the other experiments because it di d not entail shifting attention. Participants had to respond only based upon the stimulus at the cued location (Hopfinger, Buonocore,& Mangun, 2000). Unsurprisingly, researchers found dorsal activ ity rather than ventral activity. During the cue period, participants showed in creased activity in the IPS, SPL FEF, and MFG, which are all components of the dorsal network (Hopf inger, Buonocore,& Mangun, 2000). Perhaps participants in this study had this strong of a dorsal response because they knew they would not have to reorient attention and could thus fixate at the cued location with confidence. Targets alone evoked dorsal (SPL), ventral activity (IFG), and a region which may link the two, the MFG (Hopfinger, Buonocore,& Mangun, 2000). Another experiment, using a modified Poser paradigm, presented par ticipants with a cue that did not provide any inform ation about target location (L epsien & Pollman, 2002). It was intended to capture stimulus-driven attention on ly. Thus, results are not contaminated by goaldriven spatial expectation. The cues were either valid or invalid (Lepsien & Pollman, 2002).The researchers found that for both va lid and invalid trials, the follo wing regions were active: the FEF, MFG, aI, SPL, and TPJ (Lepsien & Pollm an, 2002). This pattern suggests a general
40 stimulus-driven network composed of both dor sal and ventral regions The only region that showed a main effect of validity was the MFG, which may be a mediator between networks (Corbetta & Shulman, 2008). Give n participants could not create spatial orienti ng expectations, it would have been interesti ng to see what was going on dur ing cue presentation if the researchers used event-related fMRI. The experiments discussed earlier have only shown dorsal activity while participants are focusing on a stationary fixation point or duri ng presentation of a stationary cue. One experiment, however, found dorsal activity using a visual search design (de Fockert, Rees, Frith, & Lavie, 2004). In this unique experiment, there was no explicit cue to indicate location (de Fockert, Rees, Frith, & Lavie, 2004). Without a cu e to create spatial expe ctations, participants thus could not have an expectation mismatch (de Fockert, Rees, Frith, & Lavie, 2004). Participants had to indicate the orientation of a line within a gr een circular target though they did not know where it would occur (de Fockert, R ees, Frith, & Lavie, 2004). In some conditions of the experiment, the target circle was red, and in some conditions a distracter diamond was red (de Fockert, Rees, Frith, & Lavie, 2004). The usage of red and green targets makes the display highly visually salient. Participants thus must filter out irrelevant distracters but do not have any spatial expectation. The contrast of presence > absence of the co lored distracter was a ssociated with activity in the SPL and a region near the FEF, two region s of the dorsal system (de Fockert, Rees, Frith, & Lavie, 2004). The results provide more suppor t for the hypothesis that the dorsal system is responsible for maintaining task-relevant goal-driv en signals and filtering out distracters. The experiment also involved monitori ng eye movement of participants.
41 A different experiment suggests that this reorienting response is not limited to the visual modality. Participants were given either an auditory or visual cue, and were presented with visual and auditory stimuli simultaneou sly (Downar, Crawley, Mikulis, Da vis, 2001). Researchers first presented participants with a buzzing sound and the visual image of a square. After a varying interval the square rotated or the pitch changed by five perc ent (Downar, Crawley, Mikulis, Davis, 2001). Once again the cues could have been e ither relevant or irrelevant (i.e. a visual cue was followed by the change in orientation of the square) (Downar, Crawley, Mikulis, Davis, 2001). Several regions were common to both auditory and visual stimuli for both validity contrasts (invalid > valid), suggesting a common ne twork when participants are given an explicit sensory expectation (Downar, Crawley, Mikulis, Davis, 2001). These ventral regions were the TPJ and left aI (Downar, Crawley, Mikulis, Davi s, 2001). The presence of ventral activity only suggests stimulus-driven attention shifts may be mediated by the ventral network only when participants have to shift attention between modalit ies. When validity is not part of the analysis, both dorsal and ventral regions are active (IFG, MFG, aI, and IPS) (Downar, Crawley, Mikulis, Davis, 2001). In conclusion, current neuroimaging evid ence supports the hypothe sis that spatially directed goal-driven attention is mediated by the dorsal network. Stimulus-driven task-relevant attention, however, is mediated by both networks in the case of. The relationship between taskirrelevant salient stimuli and atten tional networks is very unclear. With an understanding of mindfulness, how to measure it, its cognitive mechanisms, the clinical significance of mindfulness, and the neural correlates of the attentional changes associated with it, it is possible to crea te a hypothesis about th e relationship between dispositional mindfulness and ne ural activity in attentional networks. I hypothesize that
42 concentrative attention is analogous to the goal-driven attention, a nd thus concentrative attention mindfulness scores will correlate positively with activity the magnitude of activity in dorsal frontoparietal network (i.e. the SPL, FEF, and IPS). I also hypothesize that receptive attention is analogous to stimulus-driven attention, and thus th at receptive attention mindfulness scores will correlate positively with activity in the ventral fr ontoparietal network (i.e. the TPJ, IFT, and aI) and the dorsal frontoparietal network. The da ta on mindfulness-based therapies mindfulness skills help alleviate symptoms of depression, a nd thus I hypothesize that the same relationships will hold for the depressed participants. Methods Overall Considerations and Overview The data used for this experiment is archiv al data. The information below is an outline of the design of the study and how the data were collected. To provide a brief overview of the procedure, all participants took a battery of instruments including the KIMS, and then within the week they completed the selective visual attention scanner in the fMRI scanner. Participants Participants ranged in age from 20 to 48, incl usive, and fell into ei ther the category of patients with Major Depressive Disorder (n = 5, four females and one male ) or healthy controls (n = 6, five females and one male). Depressed patients were recruited from ongoing studies of depression in the Department of Psychiatr y, Vanderbilt University and through ComCast television advertisements. Controls were recruited from the Vanderbilt Hospital Database of volunteer subjects. Anyone who goes to Vanderbi lt Hospital has the opportunity to become a part of the database and experimenters may contact people in this database to serve as healthy
43 controls at any point in the fu ture. All participants provided informed consent approved by the Vanderbilt University Institutional Review Board. Depressed participants met several inclusion criteria. They had to have a diagnosis of major depressive disorder (MDD) according to the DSM-IV criteria, a minimum score of 16 or above on the first 17 items of the clinician-ra ted Hamilton Depression Rating Scale (HRSD), a 20 or above on the self-report Beck Depression Inventory, and they had to be legally competent and willing to give informed consent. Participants did not meet any of the follo wing exclusion criteria. They did not have a history of bipolar affective disord er, psychosis, any other Axis I di sorders, a history of substance dependence in the past six months, personality di sorders, or subnormal in tellectual potential (IQ below 85). Women who were preg nant or planning to become pregnant were also excluded. Participants were not on antid epressant medication or on catecholaminergic antihypertensive medication, however, they were not excluded based on previous antidepressant medication usage. In addition, participants did not have curr ent suicide risk sufficien t to preclude treatment on an outpatient basis (e.g., stated intent and a formulated plan). Prospective matched control participants did not meet criteria for personality disorders, any current Axis I disorder except simple phobias, or have HRSD greater than 6 or BDI less than 10. Participants were excluded if there was eviden ce of chronic disease (s uch as cardiovascular disease or neurological disord er), obesity, or claustrophobia. All clinical interviews were conducted by Dr. Merida Grant at the Village at Vanderbilt in the Department of Psychiatry. Procedure Upon identifying a potential par ticipant, the researchers cond ucted a phone screen to rule out general medical conditions. When the partic ipant came in, Dr. Grant conducted the SCID and
44 HRSD with him or her. The KIMS and BDI ar e self-report measures, which the participant completed in the same sitting. Because the HRSD is a time-sensitive instrument, all participants came in for the brain scan within a week of completing the instruments. Task The participant completed a modified flanker task in the MRI scanner. The instructions were presented using on a high-reso lution display screen attached to the head coil. The stimuli and instructions were presented using E-Prime software (PST, Pittsburgh, PA). Participants identified the sex of the centr al face in an array of three faces by pressing an arbitrary predetermined button, the index or middle key. Th ere were three difficulty levels. The most difficult was the incongruent condition, in which the central face was flanked by faces of the opposite sex (e.g., FMF). The intermediate difficulty level was the congruent condition, in which the central face was flanked by faces of the sa me sex (e.g., FFF). In the third and easiest condition, the baseline condition, there was only a single face surrounded by a fixation cross on each side. In addition, each array had a valence, which was happy, sad, or neutral. There were nine possible stimulus conditions, with three leve ls of valence (happy, sad, and neutral) and three levels of difficulty (hard, intermediate, and easy). There was a semi-structured presentation of the faces in each session. Each session was composed of six runs, with the first and sixth runs being neutra l (N) valence. There were two different sequences of runs, with the positive (P ) and negative (S) images occupying different positions, but with each valence conditi on adjacent to itself (i.e. either N, P, P, S, S, N or N, S, S, P, P, N). This scheme of ordering was us ed to minimize mood induction. Each run was composed of twelve blocks. One block was compos ed of nine trials, with each trial being the
45 presentation of a single array. Each participan t completed one session, and the two sequences of runs were counterbalanced across participants. Image Acquisition The imaging component of the study took plac e in the Philips Intera Achieva 3T MRI scanners at Vanderbilt University Institute of Imaging Science (VUIIS). To facilitate spatial normalization, the MRI scanner acquired high resolu tion structural images in the axial plane using a 3D IR prepped 3DFFE sequence (T R = 450 ms, TE=17 ms, FOV = 24 cm, slice thickness = 4 mm). Twenty-eight axial interleaved 4.0mm functiona l slices (0.5 mm skip) were acquired parallel to the AC-PC line using a grad ient echo pulse sequence (EPI) providing whole brain coverage (T2*-weighted images sensitiv e to BOLD signal changes; TR = 3000 ms, TE = 28 ms, FOV = 24 cm, flip = 90, slice thickness = 4mm). Image Analysis All image analysis was conducted with Brai nVoyager QX. Each subject had 6 EPI runs, which were collapsed together to create one data file per participant. In this preprocessing phase (relative to the statistical analysis), 3-D motion correction, slice tim e scan correction, and temporal data smoother with a high pass filter we re performed to remove linear trends. Then, the functional data were manually aligned with the anatomical data, by interpolating the image sets into 3mm3 voxels. Afterwards, the data were translated in to Talairach space, which is a standardized coordinate system to allow for av eraging images across subjects and to identity regions of activation. Data were then smoothe d with an 8mm full-width at half-maximum Gaussian kernel. It is important to note that brain imaging uses statistics differently than many other disciplines, and it is necessary to provide a brief explanation of the statisti cal analyses used here
46 to interpret the results of this experiment. A Ge neral Linear Model (GLM) is created to look at patterns of activity among multiple individuals completing the same task (de Haan, 2008). A multiple regression analysis is used to model th e time course of the signal at one specified voxel and at one point in time (the dependent variable ) (Goebel, 2008). The mode l looks at this signal as a function of the different stimulus conditions (the independent variables, also called predictors) (de Haan, 2007). Stated in other words, the voxel time course is modeled as the sum of the defined predictors with their associated beta weights. A beta weight in this context, is therefore indexing the contribution of the specified predictor in explaining the activity of the voxel at a given point in time (G oebel, 2008). Thus, a region at wh ich the beta weight is large while the participant is viewing the stimulus of in terest is interpreted to mean there is significant neural activity in that region rela tive to other regions where the beta weight is small. Statistical significance is determined by comparing the esti mated parameters, or a linear combination of them in a contrast, to zero (de Haan, 2007). This comparison is often done using a t -test (de Haan, 2007). To provide a simple example illustrating how the predictors work, consider a simple experiment in which participants see only two kinds of faces, happy and sad faces. Imagine that at time zero the participant is st aring at a blank screen, at time points one and two the participant is viewing a happy face, and at times three and four the participant is viewing a sad face. In this hypothetical experiment imagine activity at voxel X is being examined. The predictors are happy and sad faces. The model assumes that if the stim ulus is being shown, neural activity will have a defined magnitude (say one here). While the stimulus is not being shown, neural activity will have a defined magnitude of zero. The predicto r function for happy faces would then have the value zero at time point one, the value one at ti me points two and three, and the value of zero
47 again at time points three and four. For the sad fa ces, the predictor functio n would have the value zero at time point one, the value ze ro at time points one and two, and the value one at time points three and four. Thus, the way to analyze which regions are active during viewing happy faces is to select voxels which have a large beta weight associated with the happy predictor function during time points one and two. Consider the following equati on, which would be modeling the observed value of voxel X at time point one using the happy predictor function (X1), the sad predictor function (X2), and an error term e : Y = b1 X1 + b2 X2 + e From the equation, regions with significant br ain activity while the participant is viewing the happy faces will have a larger beta weight associated with X1 than associated with X2. Analyzing a correlation between a continuous variable, such as mindfulness scores, and brain activity is done with the beta weights desc ribed above. Each indi vidual has a mindfulness score and a beta weight associated with one voxel. Plotting each individuals mindfulness score against his or her beta weight associated with a given region will provide a measure of the strength and direction of the relationship betw een mindfulness scores and neural activity in one region. The final analysis of this data will result in a graph with one data point per individual, with mindfulness scores on one axis and be ta weights on the other. The Pearsons r statistic can therefore be used to analyze the relationship be tween mindfulness and neural activity in a given region for a specific task. As mentioned earlier, imaging software conducts a different t -test for each voxel. There are 100,000 voxels in the brain (Goebel, 2008). As a result, it is necessary to have methods to control for the risk of spurious correlations. One such way is to use the r -statistic itself as a criterion, and eliminating all data points which show a correlati on with mindfulness scores below
48 a pre-specified r -value. Another way to limit spurious corre lations is to create a minimum cluster size threshold. The theory behind this method is that when a group of voxels is activated together, it is more likely the activity observed is the result of true underlying neural activity rather than statistical chance. After these threshol ds are set, it is unnecessa ry to report specific statistical values for any region that was activated. In the current experiment, there are nine diffe rent predictors. As mentioned earlier, there are three levels of valence (ha ppy, sad, and neutral) and three leve ls of difficulty (incongruent, congruent, baseline). To analyze activity associated with attenti on, two different contrasts were used. These contrasts were incongruent > neutra l baseline, and neutral incongruent > neutral baseline. The incongruent condition was chosen because it was the most difficult and would require more attentional processing than the congruent condition. Th e neutral incongruent condition was chosen to examine the neural correlat es of the cognitive task in the absence of any emotional distraction. The negative incongrue nt condition was used because ignoring the negative valence to make a cognitive task requir es additional cognitive control relative to the neutral condition. The neutral baseline was subtract ed from these two incongruent conditions in order to eliminate activity associated with ma king the motor response an d the visual perception of the stimuli independent of atte ntion. In the current experiment, r -value thresholds were manipulated so the resulting p-value would be below 0.01 for the controls and patients, respectively. In addition, a minimum cluster size of 100 voxels was set.
49 Results Concentrative Attention There was only one significant positive co rrelation between the concentrative attention subscale (Act With Awareness) and the magnitude of neural activity in either the dorsal or ventral network. It was in the control group in the left IFG (Talairach coordinates -59, 10, 17). The IFG is a part of the ventral network. There was no significant activity in either attentional network with respect to the con centrative attention measure in th e patient group. This result does not fit with the hypothesized re lationship between concentrat ive attention and the dorsal network; the IFG is exclusively a part of th e ventral network. See Table 1 and Figure 1 for a summary. Receptive Attention In contrast with the concentrative attent ion measure, there were multiple sites of significant activity in both the dorsal and ventral networks co rrelated with the measure of receptive attention (the Observe subscale). Activity was found in the frontal lobe, the junction of the temporal and parietal lobes, and in a s ubcortical region. For the control group, there was significant activity in the FEF (-25, 4, 59), A CC (-3, 19, 22), IFG (47, 1, 16) and TPJ (-42, -36, 14). In the patient group, by contra st, the only significant activity was in the FEF. These results conform to the hypothesis that receptive attenti on is positively correlate d with magnitude of neural activity in both the dorsal and ventral networks. The dorsal and ventral networks together are responsible for the reorienting response. The FEF and ACC are part of the dorsal system, and the TPJ and IFG are part of the ventral system. See Tables 2 and Figures 1 and 2 for a summary of the regions activ e in each condition
50 Table 1. List of Regions Associated with Concentrative Attention Negative Contrast Neutral Contrast Region Side x y z x y z Controls Inferior Frontal L -59 10 17 Gyrus Table 2. List of Regions Associated with Receptive Attention Negative Contrast Neutral Contrast Region Side x y z x y z Controls Inferior Frontal R 47 1 16 Gyrus Temporoparietal L -42 -36 14 Junction Anterior -3 19 22 Cingulate Frontal Eye L -25 4 59 Field Patients Frontal Eye R 17 -1 50 Field
51 Figure 1. Regions Associated with Concentrat ive and Receptive Attention in Controls Figure 2. Regions Associated with Concentrat ive and Receptive Attention in Patients
52 Discussion My hypothesis about the relati onship between mindfulness an d activity in attentional networks was partially supported. I hypothesized that receptive attention scores would predict activity in regions associated w ith the stimulus-drive n attention, which were in both the dorsal and ventral networks. In healthy controls, receptive attention scores correlated positively with the magnitude of neural activity in both the dorsal (FEF, ACC) and ventral networks (IFG, TPJ, and ACC). Thus, the hypothesis was true for the contro ls. In patients, receptive attention scores correlated positively only with the magnitude of ne ural activity in the FEF, which is part of the dorsal network. I also hypothesized that concentrative attention scores would predict activity in regions associated with the goal-driven attention, whic h is the dorsal network only. Concentrative attention scores correlated positivel y with activity only in the ventral network, the IFG, in controls. Thus, the hypothesis that conc entrative attention correlates positively with neural activity in the dorsal ne twork only is false. These results support an attentional conceptualization of mindfulness, although the hypothesis about mindfulness skills and specific attentional networks was only partially supported. In this discu ssion, I will first discuss what these results mean for mindfulness therapies. The n, I will discuss finding activity in the ventral network. The task at hand varied significantly fr om tasks which typically recruit ventral activity. Afterwards, I will discuss the limitations of usi ng a network approach and discuss the limitations of the archival data I was using. Receptive attention may be th e primary mindfulness skill because receptive attention scores correlated with more regions in attentiona l networks than concentr ative attention scores did. The overarching conclusion thus is that receptive attention may be the underlying skill
53 driving the efficacy of mindfulness-based therap ies. As discussed in the introduction, a study found no difference in attentional performance betw een a group of long-term meditators trained in concentrative atte ntion styles of meditation and a group trained in recep tive attention styles of meditation (Chan & Wollacott (2007). Theoreticall y, if the skills are di stinct, any attentional differences would be magnified in long-term meditators. The l ack of difference the study found can be explained if receptive a ttention develops from concentrat ive attention as Jha, Baime, & Krompinger (2007) suggest and meditators both even tually ended up with th e same skill set. If receptive attention is the primary underlying skill a nd it eventually develops from concentrative attention, mindfulness techniques that cultivate re ceptive attention may be the most efficient way to bring about the benefits of mindfulness th erapy. One such receptive attention therapy is Vipassana meditation. In addition to directing mindfulness therapies mo re specifically, clinical trials need more rigorous methodology to be labeled empirically-s upported therapy. The primary issue is offering an active treatment control group. The majority of st udies have used a treatment as usual or a wait-listed control group. Th e treatment as usual group is not st andardized and is not a suitable control. The most conservative and ideal control group is a comparison with a proven therapy. I think a control condition using d eep breathing exercises or guided imagery exercises would be an ideal control condition. These treat ments have been shown to be effective, and mindfulness purports to be a better treatment and work w ith different mechanisms compared to most relaxation treatments. Moreover, studies need to use multiple therapists, and each therapist needs to provide therapy to only one treatment condi tion. Furthermore, well-v alidated and accepted dependent measures are necessary, such as a di agnosis using the DSM, BDI scores, BAI scores, etc. Many mindfulness therapies seem to show be nefits in the long term relative to an active
54 control treatment, such as with the depr ession relapse study and MBCT compared to antidepressant medication (Teasdale et al., 2008). Thus, addi ng a long follow-up procedure may show benefits to these therapie s that are not readily apparent. Cost-benefit analyses may also help make a stronger case for mindfulness-base d therapies. Most mindfulness therapies are designed to be conducted in a group setting, which may pose financial advantages over an individual medical treat ment or psychotherapy. The current study also found a strong correlati on between receptive attention scores and activity in both the dorsal and ventral networks. According to the literature, the ventral network responds primarily to the presentation of a beha viorally relevant stimulus in an unexpected location (Indova & Macaluso, 2007; Kincade et al., 2005; Corbetta et al. 2002; Arrington et al. 2000). Therefore, it is somewhat surprising there was significant activity in the ventral network in this current task. The stimuli here appeared co nsistently in the same lo cation, in the center of the screen. It is possible this ac tivity is the result of the inherent salienc of negative faces, which may be intrinsically behaviorally relevant to humans for evolutionary reasons. Responding appropriately and in a timely manner to situatio ns in which negative emotion is involved may have been beneficial for survival. Given the dissimilarity of the current experiment to the majority of the experiments in the literature review but still finding ventral activity raises several po ssibilities. The first possibility is that the ventral network does pl ay a role in affective processi ng, and thus it is not merely an attentional network. Two core components of the ventral network, the IFG and TPJ, were active during the presentation of these very salient negativ e faces. The way to see if it is the ventral network at play rather than the spontaneous appearance of activity in both brain regions is to use a functional connectivity an alysis. If the time course of the fMRI signal is strongly correlated
55 between two regions, compared to the correlation with one region to the rest of the brain, it supports the idea that a network is responsible for the observed phenomenon rather than both regions spontaneously being active together (Fox et al., 2006). In f act, the FC analysis performed by Fox et al. (2006), showed that the time course of activity in one region of the ventral network, the ventral frontal cortex (VFC), correlated the most strongly with the time course of activity in another region, the TPJ. The IFG, one of the re gions in which the current study found activity, is a part of the VFC. Thus, the re search by Fox et al. (2006a) suppor ts the conclusion that perhaps true ventral network activity was found during the course of the present study because the TPJ and IFG may be the most important components of the ventral network. They also found the same relationship with a pair of regions in th e dorsal network, the frontal eye field and the intraparietal sulcus. There is clearly the need for more investigation in to the function of both attentional networks, their relationship to each othe r, and the relationship of regions within each network. On the other hand, it may be useful to abandon a network approach to major cognitive processes, such as attention. Using a networ k to explain a major function may be overly reductive, because it assumes a relatively large ar ea of the brain serves the same function and we have it precisely operationalized. The key example of this is with Richard Davidsons model of cerebral asymmetry to explain mo tivation and affect (D avidson, 1992). According to this model, the left hemisphere is associated with positiv e, approach-related emotions and the right hemisphere is associated with negative, a voidance-related emotions (Davidson, 1992). This model is reductive because it devotes each w hole hemisphere to a single function. When contradictory results were found, re searchers revised the hypothesis mu ltiple times to try to make the data fit a clear left/right hemisphere model. Most of this research was conducted with EEG
56 methods, due to its extremely high temporal resolution (Sobotka, Davidson, & Seunulis, 1992; Davidson et al. 1990; Davidson, Marshall, Tomarken, & Henriques, 2000). However, much of it has been done with tasks like asking participants to bisect a line and m easuring the asymmetry of the bisection (Friedman & Forster, 2005), dich otic listening tasks (Kap rinis, 2005), and tasks with chimeric faces (Friedman & Forster, 2005). These methods are very indirect. Evidence supporting the left-ri ght hemisphere usage asymme try conclusion comes from research on facial expressions dur ing emotion-eliciting films (Dav idson et al. 1990), tasks with an approach or avoidance-based financia l game (Sobotka, Davidson, & Seunulis, 1992), emotional anticipatory responses of individu als with phobia to public speaking (Davidson, Marshall, Tomarken, & Henriques, 2000), and ev en with 72-hour old infants responding to aversive-tasting solutions (Davidson et al., 1990). The simple dichotomy presented here makes very straight-forward an d concrete predictions. However, conflicting evidence abounds. Research by Friedman and Forster (2005) suggests that anticipatory responses in the abse nce of affective arousal show an opposite pattern of lateralization. I, however, doubt that affective arousal and an ticipation can ever be truly separated. In addition, the same study by Frie dman and Forster (2005) found that when participants were given a task that primes approach-related behavior (but not avoidance), participants showed increase creativity and in creased right-hemisphere activity. Why would creativity be associated with avoidance? Moreover, a study on di chotic listening in bipolar patients during the acute mania phase showed a similar reversal of expectations from Davidsons predictions (Kaprinis et al., 1995). This theory has been continually tested and reformulated to try and make sense of all the incongruities in the data, but it probably will be unable to do so because grouping the brain into halves and saying each half has a single, unified, and consistent
57 function is a gross oversimplification. One exam ple of seemingly desperate reorganization is from a current study in press (Koch, Holland, & van Knippenberg) saying the inconsistent findings can be explained by controlling how diffuse versus concrete the stimuli and affective inductions are while st ill admitting concrete stimuli can induce diffuse states. Thus, the network approach to cognition has major limita tions. An alternative to the network approach is to look at brain regions sepa rately, without postula ting there are many regions serving a single, specific, and operationalized function. The limita tion to this approach though is that brain imaging risks becoming the new phrenology, mean ing it assigns a single function to a single region rather than looking at interactions Though we do not understand the relationship among different neural regions, they are never truly independent. Because this study used archival data, it is necessary to discuss the limitations inherent to the design. Firstly, there we re only 11 participants in total, wi th five depressed patients and six healthy controls. The small number of participants greatly limits the statistical power. Also, the use of emotional faces in the stimulus array ma y limit the interpretati on of the results for a selective visual attention task. It is difficult to say which activity is the result of emotional face processing and which activity is the result of attentional contro l processes. This potential confound is particularly salient in the negative incongruent > ne utral baseline contrast. The majority of the activity found was using this ne gative incongruent > neut ral baseline contrast. The selective visual attention tasks discusse d in the literature review used simple geometric shapes, and varied parameters like cont rast and luminance. If I could design the study, I would have used a simple stimulus array like the one used by Kim et al. (1999) that does not contain any emotionally salient features. My id eal experiment would follow the general pattern of experiments in the literature review, the Posn er paradigm. I would give the participants the
58 KIMS, then have them complete the central expe ctancy task used by Kim et al. (1999). In this task, participants were given a cu e that predicted the location of the target accurately 80 percent of the time, and predicted it in correctly 20 percent of the time (Kim et al., 1999). They had to respond by pushing a button when th e target was an X but not a cross (+) (Kim et al., 1999). Furthermore, the experiments which show differen tial activity in the dorsal and ventral networks use event-related designs, meaning they acquire images during both an anticipation period and a target presentation period. Thus I would use an event-related design to improve the experiment and monitor eye movements. In addition to changing the experimental tas k, I would use a repeated measures design and actually teach the particip ants mindfulness skills and use a between-group comparison. I would also have a control group which would receive relaxation training or some variant of attentional training. I would give the participants the Five Fact or Mindfulness Questionnaire, an improved version of the KIMS. Furthermore, I would have them complete the attention task in the scanner before and after the mindfulness trai ning. This way, it would be possible to study the mindfulness therapy itself, track changes in mindfulness skills, and examine the relationship between mindfulness and neural ac tivity in attentional networks in a true selective visual attention task. It would be in teresting to see if improvements in symptoms correlate with an increase in activity in attentional brain areas. In conclusion, though the relationship be tween mindfulness and neural activity in attentional networks is still unclear, there is a strong relationship between mindfulness scores and the magnitude of neural activity in both attent ional networks. Just because the evidence did not show the predicted relationship between the two t ypes of mindfulness skills and the attentional networks, it still supports the conclusion that mindfulness is in trinsically tied to attentional
59 processes. Indeed, with the modifications in e xperimental design proposed above and addition of functional connectivity analyses, th ere is fertile ground for future exploration of the relationship between mindfulness and attention References American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders (Revised 4th Ed). Washington, D.C.: Author Anderson, N.D., Lau, M.A., Segal, Z.V., & Bishop, S.R. (2007). Mindfulness-based stress reduction and attentional control. Clinical Psychology and Psychotherapy, 14, 449-465 Arch, J.J., & Craske, M.G. (2006). Mechanisms of mindfulness: Emotion regulation following a focused breathing induction. Behavior Research and Therapy, 44, 1849-1858 Arrington, CM., Carr, T.H., Mayer, A.R., & Rao, S.M. (2000). Neural mechanisms of visual attention: Object-based sele ction of a region in space. Journal of Cognitive Neuroscience, 12, 106-117 Baer, R.A. (2003). Mindfulness training as a clin ical intervention: A c onceptual and empirical review. Clinical Psychology: Science and Practice, 10(2), 125-143 Baer, R.A., Fischer, S.F., & Huss, D.B. (2006). Mindfulness and acceptance in the treatment of disordered eating. Journal of Rational-Emotive & C ognitive-Behavior Therapy, 23(4), 281-300 Baer, R.A., Smith, G.T, & Allen, K.B. (2004). Assessment of mindfulness by self-report: The Kentucky inventory of mindfulness skills. Assessment,11(3), 191-206 Baer, R.A., Smith, G.T., Hopkins, J., Krietemeye r, J., & Toney, L. (2006). Using self-report assessment methods to explore facets of mindfulness. Assessment, 13(1), 27-45 Bishop et al. (2004). Mindfulness: A proposed operational definition. Clinical Psychology: Science and Practice, 11, 230-241 Bond, F.W, & Bunce, D. (2000). Mediators of change in emotion-focused and problemfocused worksite stress management interventions. Journal of Occupational Health Psychology, 5(1), 156-163 Brown, K.W., Ryan, R.M., & Creswell, J.D. (2 007). Mindfulness: Theore tical foundations and evidence for its salutary effects. Psychological Inquiry,18(4), 211-237 Cahn, B.R. & Polich, J. (2006). Me ditation states and traits: EEG, ERP, and neuroimaging studies. Psychological Bulletin, 132(2), 180-211
60 Chambers, R. Chuen Yee Lo, B., & Allen, N.B. (2008). The impact of intensive mindfulness training on attentional control, cognitive style, and affect. Cognitive Therapy Research, 32, 303-322 Chan, D., & Wollacott, M. (2007). Effects of level of meditation experience on attentional focus: Is the efficiency of executiv e or orientation networks improved? The Journal of Alternative and Compleme ntary Medicine, 13(6), 651-657 Corbetta, M., Kincade, J.M, Ollinger, J.M., McAvoy, M.P., Shulman, G.L. (2000). Voluntary orienting is dissociated from target dete ction in human posteri or parietal cortex. Nature Neuroscience, 3(3) 297 Corbetta, M., Kincade, J.M, Shulman, G.L. ( 2002). Neural systems for visual orienting and their relationships to spatial working memory. Journal of Cognitive Neuroscience, 14(3), 508523 Corbetta, M., Patel, G., & Shulman, G.L. (2008). The reorienting system of the human brain: From environment to theory of mind. Neuron, 58, 306-324 de Fockert, J., Rees, G., Frith, C., & Lavie, N. (2004). Neural correlates of attentional capture in visual search. Journal of Cognitive Neuroscience, 16(5), 751-759 Desimone, R., Duncan, J. (1995). Neural m echanisms of selective visual attention. Annual Review of Neuroscience, 18, 193-222 Downar, J., Crawley, A.P., Mikulis, D.J., Davis, K.D. (2001). The effect of task relevance on cortical response to changes in visual and auditory stimuli: An event-related fMRI study. Neuroimage, 14, 1256-1267 effectiveness trial. Behavior Modification, 31(4), 488-511 Fan, J., McCandliss, B.D., Fosella, J., Flombaum, J.I., & Posner, M.I. (2 005). The activation of attentional networks. Neuroimage, 26, 471-479. Fox, M.D., Corbetta, M., Snyder, A.Z., Vincent, J.L, & Raichle, M.E. (2006). Spontaneous neuronal activity distinguishes human dor sal and ventral attention systems. Proceedings of the National Academy of Sciences, 103(26), 10046-10051 Fox, M.D., Snyder, A.Z., Zacks, J.M., & Raichle, M.E. (2006). Coherent Spontaneous Activity Accounts for trial to trial variabil ity in human evoked brain responses. Nature Neuroscience, 9(1), 23-25 Goldstein, B. (2005). Cognitive Psychology. Belmont: Wadsworth. Hayes, S.C. (2004). Acceptance and commitment therapy, relational frame theory, and the third wave of behavioral and cognitive therapies, Behavior Therapy, 35 639-665
61 Hayes, S.C. (2004). Operationalizing mi ndfuness without unnece ssary attachments. Clinical Psychology: Science and Practice, 11 249-254 Hopfinger, J.B., Buonocore, M.H., & Mangun, G. R. (2000). The neural mechanisms of topdown attentional control. Nature Neuroscience, 3(3), 284-291 Indova, I., & Macaluso, E. (2007) Dissociation of stimulus re levance and saliency factors during shifts of visuospatial attention. Cerebral Cortex, 17, 1701-1711 Ivanovski B, Malhi GS. (2007).The psychological and neurophysiological concomitants of mindfulness forms of meditation. Acta Neuropsychiatrica 19, 76-91 Kabat-zinn, J., Wheeler, E., Light T., Skillings, A., Scharf, M., Cropley, T.G., Hosmer, D., Bernhard, J.D. (1998). Influence of a mi ndfulness meditation-based stress reduction intervention on rates of skin clearing in patients with moderate to severe psoriasis undergoing phototherapy (UVB) and phot ochemotherapy (PUVA). Psychosomatic Medicine, 60, 625-632 Kincade, J.M., Abrams, R.A., Astafiev, S.V., Shulman, G.L., & Corbetta, M. (2005). An eventrelated functional magnetic resonance imag ing study of voluntary and stimulus-driven orienting of attention. Journal of Neuroscience, 25(18), 4593-4604 Lappalainen, R., Lehtonen, T., Skarp, E., Taubert, E., Ojanen, M., & Hayes, S.C. (2007). The impact of CBT and ACT models using psychology trainee therapists a preliminary controlled Leary, M.R, & Tate, E.B. (2007). The multi-faceted nature of mindfulness. Psychological Inquiry, 18(4), 251-25 Lepsien, J., & Pollman, S. (2002). Covert reorien ting and inhibition of re turn: An event-related fMRI study. Journal of Cognitive Neuroscience, 14(2), 127-144 Lundgreen, T., Dahl, J., Melin, L., & Kies, B. (2006). Evaluation of Acceptance and Commitment Therapy for Drug Refractory Epilep sy: A Randomized Controlled Trial in South AfricaA Pilot Study. Epilepsia, 47(12), 2173-2179 Marin, J.R. (1997). Mindfulne ss: A proposed common factor. Journal of Psychotherapeutic Integration, 7 (4), 291-312 McKee, L., Zvolensky, M.J., Solomon, S.E., Bernstein, A., & Leen-Feldner, E. (2007). Emotional-vulnerability and mindfulness: A prel iminary test of associations among negative affectivity, anxiety sensitivity, and mindfulness skills. Cognitive Behavior Therapy, 36(2), 91101 Moore, A., & Malinowski, P. (2009). Meditatio n, mindfulness, and cognitive flexibility. Consciousness and Cognition, 18, 176-186 Ortner, C.N., Sachne, J., Kilner, E., & Zela nzo, P.D. Mindfulness meditation and reduced emotional interference on a cognitive task. (2007). Motivation and Emotion, 31, 271-283
62 Peelen, M.V., Heslenfeld, D.J., & Theeuwes, J. (2004). Endogenous and exogenous attention shifts are mediated by the same large-scale neural network. Neuroimage, 22, 822-830 Pessoa, L., Kastner, S., Ungeleider, L.G. (2002). Attentional control of the processing of neutral and emotional stimuli. Cognitive Brain Research, 15, 31-45 Pessoa, L., Kastner, S., Ungerleider, L.G. (2003). Neuroimaging studies of attention: From modulation of sensory proces sing to top-down control. Journal of Neuroscience, 23(10) 39903998 Ramel, W., Goldin, P.R., Carmona, P.E., & Mc Quoid, J.R. (2004). The effects of mindfulness meditation on cognitive processes and a ffect in patients with past depression. Cognitive Therapy and Research, 28(4), 433-455 Shapiro, S.L., Schwartz, G.E., & Bonner, G. (1998). Effects of mi ndfulness-based stress reduction on medical an d premedical students. Journal of Behavi oral Medicine, 21(6) 581-599 Slagter, H.A., Lutz, A., Greischar, L.L., Fr ancis, A.D., Neiuwenhuis, S., Davis, J.M., & Davidson, R.J. (2007). Mental training affects distribution of limited brain resources. Public Library of Sciences Biology, 5(1228), 138 Smith, B.W., Shelley, B.M, Dalen, J., Wiggins, K., & Tooley, E. (2008). A pilot study comparing the effects of mindfulness-based and co gnitive-behavioral stress reduction. The Journal of Alternative and Compleme ntary Medicine, 14(3), 251-258 Speca, M., Carlson, L.E., Goodey, E., & Angen, M. (2000). A randomized, wait-list controlled clinical trial: The effect of a mindfulness meditation-based stress reduction program on mood and symptoms of stress in cancer outpatients. Psychosomatic Medicine, 62, 613-622 Tan, Y. et al. (2007). Short-term meditation tr aining improves attenti on and self-regulation. Proceedings of the natural academy of sciences, 104(43), 17152-17156 Teasdale, J.D., Byford, S., Tayl or, R.S., Watkins, E., Holden, E., White, K., Barrett, B., Byng, R., Evans, A., & Mullan, E. (2008). Mindfulness-based cognitive therapy to prevent relapse in recurrent depression. Journal of Consulting and Clinical Psychology, 76(6), 966-978 Teasdale, J.D., Segal, Z.V., Williams, J. M., Ridegeway, V.A., Soulsby, J.M. (2000). Prevention of relapse/recurrence in major de pression by mindfulness-ba sed cognitive therapy. Journal of Consulting and Clinical Psychology, 68(4), 615-623 Thompson, B.L. & Waltz, J. (2007). Everyda y mindfulness and mindfulness meditation: Overlapping constructs or not? Personality and Individua l Differences, 43, 1875-1885 Valentine, E.R., Sweet, P.L.G. (1999). Meditation and attention: A comparison of the effects of concentrative and mindfulness meditation on sustained attention. Mental Health, Religion, & Culture, 2(1), 651-657
63 Vujanovic, A.A., Youngwirth, N.E., Johnson, K. A., & Zvolensky, M.J. (2009). Mindfulnessbased acceptance and post-traumatic stress symptoms among trauma-exposed adults without axis 1 psychopathology. Journal of Anxiety Disorders, 23, 297-303 Wenk-Sormaz, H. (2003). Meditation can reduce habitual responding. Advances, 21(3), 33-49 William, J.M, Teasdale, J.D., & Segal, Z.V. (2000). Mindfulness-based cognitive therapy reduces overgeneral autobi ographical memory in form erly depressed patients. Journal of Abnormal Psychology, 109(1) 150-155 Williams, J.M. et al. (2008). Mindfulness-based cognitive therapy (MBCT) in bipolar disorder: Preliminary evaluation of immediate effects on between-episode functioning. Journal of Affective Disorders, 107 275-279