EFFECTS OF MEDITATION TRAINING ON ATTENTIONAL NETWORKS: A RANDOMIZED CONTROLLED TRIAL EXAMINING PSYCHOMETRIC AND ELECTROPHYSIOLOGICAL (EEG) MEASURES by ADITI A JOSHI A DISSERTATION Presented to the Department of Human Physiology and the Graduate School of the University of Oregon in partial fulfillment of the requirements for the degree of Doctor of Philosophy December 2007 11 "Effects of Meditation Training on Attentional Networks: A Randomized Controlled Trial Examining Psychometric and Electrophysiological (EEG) Measures," a dissertation prepared by Aditi A Joshi in partial fulflllment of the requirements for the Doctor of Philosophy degree in the Department of Human Physiology. This dissertation has been approved and accepted by: Dr. Mm.jofr'e Woollacott, Chair ofthe Examining Committee /J / Date ) 1/)'v!/J?1~L/~ ;?C;ll? / Committee in Charge: Dr. Marjorie Woollacott, Chair Dr. Paul van Donkelaar Dr. Michael Anderson Dr. Li-Shan Chou Accepted by: Dean of the Graduate School iii An Abstract of the Dissertation of Aditi A Joshi for the degree of Doctor of Philosophy in the Department of Human Physiology to be taken December 2007 Title: EFFECTS OF MEDITATION TRAINING ON ATTENTIONAL NETWORKS: A RANDOMIZED CONTROLLED TRIAL EXAMINING PSYCHOMETRIC AND ELECTROPHYSIOLOGICAL (EEG) MEASURES Approved: _ Dr. Marjorie Woollacott Meditation has been defined as a "group ofpractices that self-regulate the body and mind, thereby affecting mental events by engaging a specific attentional set" (Cahn & Polich, 2006). We conducted a randomized, longitudinal trial to examine the effects of concentrative meditation training (40 min/day, 5 days/week for 8 weeks) on top-down, voluntary control of attention with a progressive muscle relaxation training group as a control. To determine if training produced changes in attentional network efficiency we compared, pre- and post-training, mean validity effect scores (difference between invalid cue and center cue reaction time) in the contingent capture paradigm (Folk et aI., 1992). The meditation group showed a trend towards improvement of top-down attention while the relaxation group did not. IV Using EEG we assessed the changes in amplitudes of wavelets during periods of mind-wandering and meditation. Periods in which subjects were on- vs. off-focus during the meditation task were identified by asking subjects to make button presses whenever the mind wandered and also at probe tones, if they were off-focus. After training, the episodes of mind-wandering were significantly lower in the meditation group as compared to the relaxation group. Increased amplitudes of alpha and theta EEG frequencies in the occipital and right parietal areas were seen during the meditation task for the meditation but not the relaxation group as an effect of training. A baseline EEG trait effect of reduced mental activity was seen (meditation training: occipital and right parietal areas; relaxation training: only occipital areas). Within a given meditation session, prior to training, alpha and theta activity was lower in on-focus conditions (occurring immediately after subjects discovered they were off-focus and returned to active focus on the breath/syllable) compared to meditative focus segments. After training, we found higher alpha amplitude in periods of meditative focus as compared to periods of mind wandering for both groups. However, the meditation group showed significantly higher theta amplitude than the relaxation group during the meditative state segments. v CURRICULUM VITAE NAME OF AUTHOR: Aditi A Joshi GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon, Eugene, OR College of Ayurved, Nasik (Affiliated to University ofPune, Pune, India) G.S.G. College of Ayurved, Ahmednagar, (Affiliated to University ofPune, Pune, India) DEGREES AWARDED: Doctor of Philosophy, Department of Human Physiology, University of Oregon, December 2007 Master of Ayurvedic Medicine, College of Ayurved, Nasik, March 2000 Bachelor ofAyurvedic Medicine and Surgery, G.S.G. College of Ayurved, Ahmednagar, October 1994 AREAS OF SPECIAL INTEREST: Cognitive Neuroscience, Meditation, Alternative and Complementary Medicine PROFESSIONAL EXPERIENCE: Graduate Research Assistant: Woollacott lab, Human Physiology Department (2004- till present) Graduate Teaching Assistant: Human Physiology Department, Vi Ayurveda (Traditional Indian medicine) Consultant: Self Employed Professional, 2001- Sept. 2004 Associate Consultant, Dr. Jagadisha Memorial Trust. (An Organization for working with adolescents) Housemanship: Ayurvedhospital, Nashik, Dec. 1996-Nov. 1997 Administration of day to day therapeutic management to patients. Internships: Intensive Care Unit at Ayurved Hospital Nashik, June 1996 - Nov. 1996 Assist in Emergency Management of Patients, maintain various registers. Ayurved Hospital, Ahmednagar, Feb. 1995 - Jan. 1996 Rotating Intern-ship in various departments like Medicine, Surgery, Gynecology and Obstetrics. GRANTS, AWARDS AND HONORS: Mind and Life summer research institute scholarship 2006 and 2007. Watumull Foundation Scholarship, 2004 and 2007. Sushil Jajodia Scholarship by University of Oregon, 2005-06. Travel Grant Scholarship from Sakal India Foundation in July 2004. University top Rank at MD (Kayachikitsa) Final. "Godutai Kashikar" prize for securing fIrst position amongst women candidates in the college at Final BAMS Exam. "Dr. K.V.Joshi, Zanjibar" prize and "Hari Anant Kulkarni" prize in RasaShastra Bhaishajya Kalpana (Formulation of Herbal Drugs). vii ACKNOWLEDGMENTS I wish to express my sincere gratitude to Professor Marjorie Woollacott who has mentored my doctoral training, for facilitating this study. I thank her for guiding me at every phase of the study. I extend my thanks to Dr. Michael Anderson who shared his expertise in designing and conducting the present study. I am thankful to Dr. Paul van Donkelaar and Dr. Li Shan Chou for their guidance and effort as my committee members. This study is a team effort; I thank Dr. Don Tucker and Dr. Tom Dishon for collaboration and allowing use oflab facilities. Without co-operation from other graduate students like Benjamin Levy, Tarik S. Bel-Bahar, Teresa Hawkes and Sandra Saavedra, the study would not be possible. Sincere effort from Emily Longbrake and technological support from Cooper Boydston and Wayne Maneslle has helped the study immensely. I also thank my labmates for being kind and supportive. My family, especially my mother, has been a source of inspiration. My dear friends Ms. Divya Bhasin and Ms. Radhika Naik deserve special thanks. Last but not least I thank all the near and dear ones who have been with me during this learning phase. V111 TABLE OF CONTENTS Chapter Page 1. INTRODUCTION 1 II. MEDITATION TRAINING AND IMPROVEMENT OF THE ABILITY TO OVERRIDE ATTENTIONAL CAPTURE......................................................... 22 III. A RANDOMIZED CONTROLLED TRIAL COMPARING CONCENTRATIVE MEDITATION VS. RELAXATION TRAINING: IMPACT ON MENTAL FOCUS AND EEG CHARACTERISTICS.................. 43 IV. ATTENTIONAL REGULATION IN MEDITATION: COMPARISON OF EEG CHARACTERISTICS DURING PERIODS OF ON- VS OFF-FOCUS... 77 V. GENERAL DISCUSSION 101 APPENDICES A. MEDITATION INSTRUCTIONS 107 B. LIFESTYLE QUESTIONNAIRE 113 C. EEG SESSION INSTRUCTION 123 BIBLIOGRAPHy................................................................................................................... 126 IX LIST OF FIGURES Figure Page 1. Task display for contingent capture paradigm 32 2. The effect of cuing as a function of training and group................................. 37 3. Occipital electrode cluster.............................................................................. 58 4. Parietal electrode cluster 59 5. Frontal electrode cluster...... 60 6a.Wavelets from frontal electrode cluster in focused attention state, prior to training " . 61 6b Wavelets from frontal electrode cluster in focused attention state, after training........................................................................................................... 62 7. Effect ofmeditation vs. relaxation training on normalized theta frequency amplitude..... 64 8. Effect of meditation vs. relaxation training on normalized alpha Hz amplitude for occipital and right parietal electrode clusters. 65 9. Effect of meditation vs. relaxation training on baseline alpha Hz amplitude for occipital electrode clusters. 67 10. Changes in mean alpha amplitude as a function of condition in the frontal electrode cluster prior to training. 86 11. Changes in mean alpha amplitude as a function of condition in the occipital electrode cluster prior to training 87 12. Changes in mean alpha amplitude as a function of condition in the parietal electrode cluster prior to training 88 13. Changes in mean theta amplitude as a function of condition in the occipital electrode cluster prior to training 89 14. Changes in mean alpha amplitude as a function of condition in the frontal electrode cluster after training 92 15. Alpha amplitude as a function ofmental condition in the parietal electrodecluster after training 93 16. Changes in mean theta amplitude as a function of condition in the frontal electrode cluster 94 x LIST OF TABLES Table Page 1. Demographics of the participants in the study...... 29 2. Accuracy as a function of Target-Property condition and Training 34 3. Reaction times (in ms) as a function of training and cue conditions........ 35 4. Cost of invalid cue as a function of target-property condition 36 5. Demographics of the participants 51 6. Button press responses 56 7. Change in alpha and theta amplitudes as a function of time.. 63 8. Change in resting state EEG amplitude as a function of training for meditation and relaxation groups......................................................................... 66 1 CHAPTER I INTRODUCTION Healthcare and Alternative & Complementary Medicine Alternative and complementary medicine has carved a niche for itself in the health care system of the U.S. and most other countries for a variety of reasons. For example, its cost-effectiveness, its ability to provide solutions to some challenges unmet by western biomedicine, such as integrating care of the mind, body and emotions of an individual, and its ease of administration have made alternative and complementary medicine increasingly popular. Mind and body medicine and, in particular, meditation is being widely practiced (see Kabat-Zion, 2003 for a review). Though meditation, or the practice of focusing and relaxing the mind, has been traditionally considered an integral activity in various religious traditions, recent research has suggested that it is helpful in improving emotional well-being and reducing stress, and thus meditation is now a widely practiced psychological intervention used in clinical settings for generating relaxation responses (Benson, Greenwood, & Klemchuk, 1975). Various types of meditation, including concentrative meditation and mindfulness meditation, have been successfully employed in patient populations suffering from a variety of diseases related to or modulated by stress levels. Hypertension, alcohol intake, substance abuse, psoriasis and headaches are some disorders where mindfulness meditation or "the relaxation response practice" (non-religious medical based meditative 2 practice developed by Dr. Benson) has shown beneficial effects (Benson, Greenwood, & Klemchuk, 1975; Kabat-Zinn et aI., 1998). We will now explore definitions of meditation that can accommodate various techniques of meditation and are applicable to the present study. Meditation- Defining Characteristics The word 'meditation' is derived from the Latin root 'meditari-' which means to'contemplate or reflect.' Since meditation encompasses a variety of tasks, defining meditation is a challenge. However, meditation techniques may be broadly classified along a continuum, based on the attentional processes involved, with the two poles being identified as concentrative meditation and mindfulness meditation, depending on how attentional processes are employed. Mindfulness meditation focuses on awareness of thoughts and sensations, as a non-attached observer, without judgment, and includes such practices as Zen and Vipassana meditation. Concentrative meditation involves focus on a specific mental or sensory activity, including a repeated bodily sensation such as the breath, or a thought, such as a mantra syllable, and includes yogic meditation and Samatha meditation, as well as aspects of Transcendental Meditation which involve repetition of a syllable (Cahn & Polich, 2006). In the sections below, defining characteristics of meditation will be discussed in terms of behavioral and physiological modulation that occurs as a result of a specific technique. 3 Meditation, a stress management technique for relaxation enhancement Practicing certain meditation techniques, including types of concentrative meditation and, mindfulness meditation, such as Transcendental Meditation (TM) or Zen meditation can produce a decrease in muscle tone, respiratory rate and heart rate of the practitioner (Benson, Greenwood, & Klemchuk, 1975). Hence, meditation has often been defined as a wakeful hypometabolic integrated response (Jevning, Wallace, & Beidebach, 1992). Benson (1975) terms meditation as one of the activities generating a relaxation response. However, there are also certain hyperactive styles of meditation, for example, Qi gong, in which excited physiological states are also sometimes achieved. As defined by Shapiro (1982), meditation is a family of techniques which have in common a conscious attempt to focus attention in a non-analytical way, and an attempt not to dwell on discursive, ruminating thought. Meditation, an altered state ofconsciousness Since meditation can bring about a change in the level of consciousness awareness, the relationship of meditation and consciousness has been of interest. Consciousness here is defined as "the neural activity underlying the state of waking awareness" (Eccles, 1994; Woollacott, 2004) Meditation is considered "a psychologically induced altered state of consciousness" (Dietrich, 2003; Vaitl et al., 2005). Meditation has thus been defined as an ancient technique which aims to gain a degree of control over various autonomous psychobiological processes (Davidson & Goleman, 1977) and 4 changes in states of consciousness associated with meditation have been measured through electroencephalographic (EEG) measures, functional Magnetic Resonance Imaging (tMRI) and other measures of brain activity. The state of consciousness associated with meditation can be voluntarily practiced and can be modified with training. Meditation, a mental training Mental faculties like cognition and attention can be trained! regulated with meditation. Thus meditation has also been defined as a, "group of practices that self­ regulate the body and mind, thereby affecting mental events by engaging a specific attentional set" (Cahn & Polich, 2006). Takahashi et ai. also endorse a similar view, using the definition, "meditation is an attainment of a restful yet fully alert physical and mental state practiced by many, as a self regulatory approach to emotion management" (Takahashi et aI., 2005). Finally, meditation has been defined as a "mental training" that brings about long term changes or trait changes in cognition and emotion (Lazar et aI., 2005; Lutz, Greischar, Rawlings, Ricard, & Davidson, 2004). A definition of meditation put forth by Lutz et ai. (2007) is comprehensive and can be applied to different meditative techniques. Meditation is characterized by three features 1) the claimed production of a distinctive and reproducible state that is phenomenally reportable; 2) the claimed relationship between that state and the development of specific traits and 3) the claimed progression 5 in the practice from the novice to the virtuoso (Lutz et aI, 2007). Similarly, Cardoso et al argue that meditation practices use a definite technique involving self-focus skills. Meditation is a self induced practice that produces both- muscle relaxation and logic relaxation (Cardoso, de Souza, Camano, & Leite, 2004). We are interested in understanding these two components - focused attention (or self-Jocus as termed by Cardoso) and relaxation and their neural correlates in the attentional networks. Research on Meditation- Background, challenges, solutions Surveys indicate that the use of meditation as an alternative and complementary medicine is increasing (Barnes, Powell-Griner, McFann, & Nahin, 2004; Wolsko, Eisenberg, Davis, & Phillips, 2004). Barnes et al reported that 62% of American individuals in a survey had used mind and body medicine within the survey year and that meditation accounted for 7.6% of mind and body medicine practiced within a year. This reflects an increase in the practice of meditation, and the increase is not only quantitative; studies have applied meditation successfully as a therapy to meet challenges like improving the mental health of HIV positive individuals, treating eating disorders and substance abuse. However, a recent NCCAM report (Ospina et aI, 2007) mentions that the physiological and neuropsychological effects of meditation practices require cautious interpretation as only a small number of studies meet the strict criteria of randomized and controlled longitudinal clinical trials, which are the hallmark of the highest levels of excellence in research studies with relation to evidence-based practice. The majority of 6 the literature compares physiological data from meditation practitioners to a control group and hence the differences between groups can also be attributed to other factors than meditation training, such as self- selection biases. A number of studies have also been limited in generalizability by the use of a waitlisted control rather than a concurrent control in a longitudinal study. Other methodological issues like non-randomization or /, inappropriate randomization, and lack of blinding reduce credibility of the studies (Ospina, 2007; Jadad, 1996). As background review, we shall now consider studies attempting to determine neural correlates of meditation and challenges faced by those studies. Thus, we consider factors that influence or should be taken into consideration while designing a research study related to meditation. Techniques used in meditation studies Selection of an appropriate technique/tool to measure state and trait effects for the meditation is crucial. The tests or tasks should be sensitive enough to measure the differences before and after training or across the two groups. Various computer-based tasks have been used to measure behavioral or physiological performance (e.g. the Stroop task, a measure of executive attentional efficiency, used by Chan and Woollacott (2007), and the Attentional Network task, a measure of executive, orienting and alerting network efficiency, used by Jha et al (2007)). These tasks typically inform us about the trait changes in meditation practitioners whereas tools like imaging give us ability to look at state changes during meditation. 7 Imaging is the most suited technique to study the spatial details of the neural substrates of meditation. Studies have reported activation in different neural substrates depending upon the meditation techniques used and the investigation techniques used, with a number of studies reporting that the frontal lobe and parietal lobe exhibit increased activation during meditative compared to baseline resting states (Dietrich, 2003; Lazar et al., 2000; Lehmann et al., 2001; Lou et al., 1999; Lou, Nowak, & Kjaer, 2005; Newberg et al., 2001; Travis, Tecce, Arenander, & Wallace, 2002). However, as meditation is a technique requiring focused internal concentration, usually in a noise-free environment, the high noise levels in magnetic resonance imaging scanners makes one cautious about the ability of individuals to enter a meditative state in this environment, and thus also about the interpretation of experimental results using this method. To help overcome this technical constraint in the use of fMRI, Lazar and colleagues attempted to accustom meditation subjects to the high noise levels in the scanner prior to testing (Lazar et al., 2000). EEG data provide the temporal resolution required for determining differences in the performance of meditation practitioners on various attention related tasks (e.g. attention blink paradigms) (Brefczynski-Lewis, Lutz, Schaefer, Levinson, & Davidson, 2007; Slagter et al., 2007). Alpha waves have an 8 to 12 Hz frequency. They are a characteristic of relaxed wakefulness. Any sensory stimulus will give rise to desynchronisation in alpha waves, also termed as an alerting response. Beta waves (13 to 30 Hz) are seen in frontal regions when a subject is engaged in intense mental activity (Chaterjee, 2000). Theta waves have a frequency of4 to 8 Hz and are common in children of age two to five.. 8 They are normally found in a state of drowsiness. Delta waves have a frequency between 0.5 to 4 Hz, are common in sleep and in infancy and in conditions like hypoxia and hypoglycemia but rare in adults. Gamma activity is at higher frequency 25 to 70 Hz and it is considered to be associated with coherence in brain activity (Knyazev, 2007). Normal EEG can be affected by age, blood glucose levels, oxygen and carbondioxide levels, temperature, sensory stimulation, during sleep, narcotics and pathological conditions. In EEGs, we assess electrical activity of (typically pyramidal cells) neurons from the cortical surface. Deeper structures like the brain stem, thalamus and hippocampus cannot be assessed significantly as their contributions to surface electrical activity are poor. EEG is a summation of activity of the synapses of the pyramidal cells, which are oriented perpendicular to the cortical surface (Chaterjee, 2000; Kandel et aI, 2004). However, with the use of dense array EEG (more than 128 electrodes) one can obtain reliable information about the source of surface recorded brainwaves. The data from studies using EEGs to measure brain activity during meditation also adds to the evidence of frontal lobe and parietal lobe activations during meditative states (Aftanas & Golocheikine, 2001; Lutz, Greischar, Rawlings, Ricard, & Davidson, 2004). However, increases shown in these studies are typically in the alpha and/or theta range. One must be cautious in comparing the results of EEG recordings and fMRI data, as increases in EEG slow wave frequencies (i.e., alpha and theta) have been interpreted by Lutz and colleagues (Lutz et aI, 2007) as a reduction or inhibition in normal activity, associated with reduced sensory, motor or cognitive processing accompanying a quiet, alert state. 9 Magnetoencephalography is used to determine source dipoles of brainwaves. Using magnetoencephalography (MEG), Yamamoto et aI. (2006) compared the current source dipoles of the brain activity in TM practitioners to a control group. The control group, who hadn't been trained in meditation techniques, performed a task of repeating a non-mantra syllable. The dipoles for the high amplitude alpha waves during TM practice were localized at the anterior cingulate cortex (ACC) and medial prefrontal cortex unlike the control group who did not exhibit high amplitude alpha waves (Yamamoto, Kitamura, Yamada, Nakashima, & Kuroda, 2006). During a meditation session the regional cerebral blood flow (rCBF) increases, which can be measured via rheoncephalography. Jevning et aI. (1996) compared the rCBF in experienced Transcendental Meditators (TM) and a rest control group. They found increased rCBF in the frontal and occipital regions in the TM group, which indicates an increase in cerebral activity in these areas during meditation; they thus concluded that meditation was an altered state of consciousness separate from sleep, since during sleep, the rCBF is decreased (Jevning, Wallace, & Beidebach, 1992). In a separate study using Photon Emission Tomography (PET) and EEG, Lou et al. (1999) identified changes in regional cerebral blood flow (rCBF) during different types of meditation involving different types of visual, kinesthetic and emotional imagery. Subjects listened to a recording which gave verbal guidance related to visualization, feelings of joy, . and experience of limb weight. Results showed that regional blood flow changes varied according to meditative content with meditation on limb weight activating parietal and superior frontal areas, whereas during the experience 10 of joy, Wernicke's area, the left parietal lobe and superior temporal lobe were activated. Finally, visualization activated the occipital lobe, except for VI. EEG analysis (20 electrodes, total) showed increases in theta frequencies throughout the meditation period (Lou et aI., 1999). Meditation may also cause changes in endogenous neurotransmitter release. Kjaer et al (2002) using Positron emission tomography (PET) scans and simultaneous EEG recordings, observed an increase in dopamine release in the ventral striatum of the basal ganglia complex during yoganidra meditation as compared to a rest condition. The increase in dopamine in the striatal area was correlated with EEG theta wave activity, and decreased blood flow to the prefrontal cortex, supporting the conclusion that the subjects were in a meditative state (Kjaer et aI., 2002; Lou, Nowak, & Kjaer, 2005). Challenges and concerns in meditation research Ospina (2007) list five major problems associated with prevIOUS meditation research in their meta-analysis of published studies in this area. Cohort studies have not taken into account self-selection biases while assembling both an exposed (meditation) group and non-exposed (control) group. The comparison or the control group was similar to the population of interest only in 21 % of published studies. In meditation training studies, lack of randomization of subjects between meditation and control groups and lack of blinding on the experimenter or participant's part, in order to curb any biases regarding expected outcome of the study, were also issues of concern. Lastly, a lack of any description of drop out rate in meditation vs. control group training studies was noted by Ospina et ai. as a weakness of many longitudinal studies. Due to these 11 drawbacks the studies had a low score on the Jadad scale, measuring the quality of clinical trials, and thus the generalizability of results from these studies was questioned. We will now give examples ofthese concerns as seen in previous studies. Population ofinterest Studies have shown high levels of within and across group variability in the population of interest, i.e. meditation practitioners, in terms of length of experience of practitioners, age and type ofpractice. Experience ofmeditator A number of studies have been conducted on experienced meditators; in order to aim for consistency across subjects in state effects, i.e., experienced subjects can reproduce a deep state ofmeditation during each testing session. For example, the studies conducted by Sim & Tsoi (1992) and Lou et al (1999) had a single group of experienced subjects. They found that experience (proficiency) in meditation changed the subjective experiences of a meditator, the physiological functioning of the neural substrates and also the anatomical patterning of the underlying structures. In a study designed to measure these differences Lo et al (2003) examined the effect of meditation on brain activity using EEG and correlated it with the subjective experience of the subjects. They state that during meditation sessions, the experienced Zen meditators (more than 11 years of practice) perceived 'inner light' during what they described as a deep stage of meditation. This was correlated with alpha blocking and the emergence of a small amplitude beta 12 rhythm. The individuals with lesser experience did not report the experience of inner light or the alpha blocking and beta activity (Lo, Huang, & Chang, 2003). A trait or long-term effect of meditation on cortical thickening was studied by Lazar (2005) in experienced meditators compared to control subjects using structural MR!. The subjects had experience in Insight or Vipassana meditation (a mindfulness based meditation). The control group had no experience in any type of meditation. The meditation group showed an increased cortical thickening in areas that have been previously associated with activation during meditative practice- the right middle frontal superior area, the prefrontal cortex and the right anterior insula. The experience in meditation was found to be positively correlated with the cortical thickening whereas the age matched controls did not exhibit cortical thickening (Lazar et aI., 2005). The above mentioned studies considered total experience of meditators in years; however, in a separate study the amount of daily practice of meditation (min/day) was found to be positively correlated with the ability to focus attention (Chan & Woollacott, 2007). Hence in designing a research study experience and proficiency both should be accounted for. The control group In randomized controlled trials, assessment of individuals to assign them to a group is important. The control group is to be matched on various criteria such as age, gender, education, occupation and lifestyle. The control group has to be assigned a comparable activity in a longitudinal study and the control group members should have 13 an equivalent training experience to that of the experimental group (equal training time, equal access to a teacher, equal belief in the validity ofthe training procedure). Waitlisted Control Group Designs: Davidson et aI, recruited 25 participants for a mindfulness based stress reduction (MBSR) seven week training program. The control group was of 16 individuals who were waitlisted for MBSR training. Self reports about positive and negative affect, anxiety and EEG changes in anterior electrode sites related to positive affect were assessed before and after training. As compared to the controls, the meditation group reported a significant decrease in negative affect. EEG recordings showed that meditators produced a greater left-sided EEG activation in central sites as compared to the controls. To study the impact of meditation on the immune system, the researchers vaccinated subjects using influenza vaccine and measured antibody titers created in response to the vaccine, both in the meditation group and a control group at the end of meditation training. In the meditation group, at the end of training there was a rise in antibody titer, which had a positive correlation with increased positive affect, which was also related to an increase in left-sided anterior activation in the brain. Thus meditation enhances immune responses (Davidson et aI., 2003). A similar waitlist control design was used by Carlson Linda et al (200 1) to study the efficacy of the Mindfulness Based Stress Reduction (MBSR) program on mood and stress management in cancer patients (Linda, Zenovia, Eileen, Maureen, & Michael, 2001). However, using a concurrent control group, instead of a wait-listed experimental group as a control, adds more credibility to the research design since the experiences are the same. 14 In a cross-sectional study Khare and Nigam (2000) compared EEG data of individuals practicing meditation and relaxation, testing subjects before, during and after a meditation/relaxation session. The meditation group showed prominent alpha activity during meditation whereas relaxers had prominent beta activity. The alpha activity persisted even at the end of the session and on opening the eyes. An unusual finding of this study was the presence inter-hemispheric symmetry in alpha activity since previous studies have shown right hemisphere dominance in alpha activity (Gaylord, 1999; Bagchi, 1957; Wallace 1970). A weakness of the study was that the meditators had a novice level experience (3 months) while the relaxation group was naIve. Thus it is difficult to compare groups with different levels of training in the two techniques (Khare & Nigam, 2000). Lutz et al. (2004) conducted a study to examine the role of neuronal synchrony as a neural mechanism underlying mental training (meditation) in long-term meditators (Tibetan Monks). A naIve control group with a different cultural background (American) and a different age range (substantially younger) was given instructions similar to loving kindness meditation a week ahead of testing. The EEGs of the long-term meditators showed high amplitude gamma frequency during meditation that was not present in the control subjects, in addition to higher level base-line gamma activity than the control group. The increase in gamma synchrony in meditators was prominently in the midfrontal and the parietotemporallobes. Lutz et al. argue that the difference in the baseline EEG characteristics between groups is not due to the difference in ages of the experimental and the control group 15 (meditation group mean age = 49 years; control group = 21 years). Even when the youngest practitioners in both the groups were compared, the ratio of gamma oscillations to the slow rhythms (4 - 13 Hz) remained higher in the meditators. A better research design would use meditators and controls that were matched as closely as possible on the basis of age, gender, cultural background and education (Lutz, Greischar, Rawlings, Ricard, & Davidson, 2004). Gaylord et al (1989) conducted a study in the black (African-American) adult population to test the effects of one year of training in Transcendental meditation (TM) techniques vs progressive muscle relaxation (PMR) or cognitive behavioral strategies (C) on EEG coherence, stress reactivity, and mental health. They found significant improvements in mental health and reduction in anxiety in the TM and PMR groups. However, only the TM trained group showed increase in alpha and theta coherence in frontal and central EEG electrodes compared to eyes closed rest. Changes were most marked in the right hemisphere. None of the groups showed longitudinal changes in EEG with training. The study design was weakened by not balancing groups for age, education, etc. In addition, EEG coherence showed a positive correlation with IQ post­ training. However, during reporting EEG coherence, the researchers grouped all the participants together, irrespective of the activity practiced. Such methods do not contribute towards refuting alternative explanations of the results (Gaylord, Orme­ Johnson, & Travis, 1989). To summarize, in studies of the efficacy of meditation training, a well-matched control group is required to increase the credibility of results. Issues like lifestyle, age, 16 gender and possibly genetics should not be ignored while matching the groups. In longitudinal studies, a careful documentation of quantitative and qualitative practice during the monitored (e.g. in class) and unmonitored (e.g. at home) phases of the training programs, with the help of logs/journals and possibly objective parameters, like Holter monitoring, is needed. Attentional networks and control of attention Meditation training can enhance the ability to attend to an object of focus and ignore distractors (Chan and Woollacott, 2007; Jha et aI, 2007). The neural processes underlying attentional control have been widely studied both physiologically and anatomically. Depending upon the specific form of information processing examined and the cortical areas facilitating the information processing, Posner and Peterson (1990) have proposed three attentional networks, including the alerting, orienting and executive control networks. Alerting has been defined as "achieving and maintaining an alert state." Research has shown that the right frontal area and parietal area are associated with the alerting function. Orienting has been defined as "selection of information from sensory input," and executive control has been defined as "resolving conflict among responses." Orienting can be viewed as attentional selection at the sensory input level whereas conflict resolution is attentional selection at the response output level. The superior parietal lobe has been hypothesized as the neural substrate for orienting and the anterior cingulate gyrus and frontal areas for executive function or conflict resolution (Fan, McCandliss, Sommer, Raz, & Posner, 2002; Posner, 1994; Posner & Rothbart, 1998). 17 Although evidence supports a degree of independence of these networks, cooperation and interdependence between the networks has also been suggested (Raz & Buh1e, 2006). As an alternative approach to attentiona1 processing models, Corbetta and Schulman have proposed a bipartite attentional network. According to their view, the dorsal attentional network consists of frontal eye fields (FEF), bilateral posterior parietal cortex (PPC), and the intraparietal sulcus (IPS). This network shows neural activations during stimulus selection, response selection and task switching, thus facilitating the endogenous, top-down, voluntary control of attention. The functions of orienting and conflict resolution would be considered part of the voluntary control of attention and hence controlled by the dorsal network of attention. The ventral system of attention consists of the temporoparietal cortex (temporoparietal junction, TPJ) and ventral frontal cortex (VFC). This system carries out stimulus driven, exogenous, bottom-up control of attention. The TPJ acts as a "circuit breaker" and helps in orienting attention to relevant stimuli (Corbetta & Shulman, 2002; Fox, Corbetta, Snyder, Vincent, & Raichle, 2006; Serences et aI., 2005). Studies have highlighted modulation of voluntary attentiona1 control via various processes like drug-effects, hypnosis and meditation (Vaitl et aI., 2005). During a focused type of meditation, the impact on the attentional networks has been hypothesized to be temporally divided into two parts, phasic control and tonic control. During phasic control, the prefrontal cortex and frontal eye fields (dorsal attention network) would lead to an initial quieting of both the mind and body. In addition, tonic control is hypothesized to be exerted by the basal gang1ia-thalamocortica11oop. These circuits are hypothesized 18 to cause alpha synchronization and maintain a state of 'restful alertness.' If an individual has 'other' experiences than transcending (e.g., awareness of environmental stimuli) it is hypothesized to result in alpha desynchronization (Travis, 2001; Travis & Wallace, 1999). In the present study we are interested in assessing the phasic control or the initial executive control of attentional processing associated with meditation. We also want to explore the trainability of the dorsal attentional network and thereby executive control and orienting. During meditation, like in any other continuous performance task participants may undergo alternating periods of diffused attention or mind-wandering and focused attention. According to Schooler's theory of mind-wandering, personally relevant goals ---------­ may be. automatically activated and make a participant drift away from a task-related goal (Smallwood, McSpadden, Luus, & Schooler, 2007; Smallwood & Schooler, 2006). It is probable that in previous research on meditation, such periods of mind-wandering were averaged in to the analysis of the data from the meditation session and thereby reduced the clarity of the results. In the present study we attempt to determine the EEG characteristics associated with on-focus vs. off-focus periods ofmeditation. 19 BRIDGE In summary, the objective of this research was to determine the effects of meditation training on the efficiency of attentional processing networks, including both short-term state effects and long term trait effects. A small number of studies have been previously performed to examine the effectiveness of meditation training on attentional processing efficiency. Though these studies give some tantalizing evidence regarding the possible improvements in processing within specific attentional networks, both as a result of short effects of meditation on an individual's state (during and immediately after meditation), and longer term effects that indicate carry over of improved processing into performance on activities of daily living, they also suffer from a number of limitations, described above. Though these initial studies have suggested that meditation can improve attentional network efficiency, the factors and neural mechanisms underlying this improvement are not clear. First, no randomized and controlled clinical trials have been performed, in which the effects on attentional networks of meditation training and a reasonably similar control training were compared. Second, though studies of meditation note that training includes both an element of relaxation and of attentional focus, no one has carefully determined the contributions of these two factors to improvement in the efficiency of attentional networks. Finally, no studies have actually dissected meditation sessions into periods of active concentration vs. periods in which the practitioner was distracted by extraneous thoughts 20 or by sleepiness. Thus, when recording such physiological correlates of meditation as EEG, it is possible that periods of different types of unfocused attention were averaged in with focused attention, thus reducing the clarity and precision of the results. To address these issues, a longitudinal pseudo-randomized controlled clinical trial was performed to examine the relative effects of concentrative meditation training vs. relaxation training on attentional network function. This allowed the comparison of the relative contributions of components of attentional concentration vs. relaxation to the neuropsychological and physiological measures of attentional network efficiency. The following tests and measurements were included in pre-test and post-tests. Neuropsychological test: The Contingent Attentional Capture task (Folk et aI, 1992; 1994; 2002) for a neuropsychological measure of the efficacy of goal-driven voluntary attentional networks vs. stimulus driven networks (dorsal vs. ventral frontal/parietal attentional systems) and personality state and trait inventories. Electroencephalograms (256 electrodes): to determine the extent to which training in relaxation vs. focused concentration + relaxation is associated with changes in the activity of brain regions associated with specific attentional networks. We hypothesized that participants trained in meditation (including mental/ cognitive relaxation) compared to relaxation alone would show significantly higher efficiency in networks of both executive and goal-driven voluntary attentional systems. In addition, we hypothesized that they would show increased alpha and theta slow wave power in sensory cortex (occipital and parietal) networks associated with decreased activity in these areas (hypothesized to be associated with reduced attention to external stimuli). We also hypothesized that there would higher 21 be levels of alpha activity during baseline rest as a result of trait effects of meditation causing subjects to remain in a more relaxed state during daily function. We also examined periods within a meditation session when practitioners were on vs. off focus on the concentration task and compared physiological characteristics (EEG) under these identified periods of differential focus. 22 CHAPTER II MEDITATION TRAINING AND IMPROVEMENT OF THE ABILITY TO OVERRIDE ATTENTIONAL CAPTURE Introduction Attention and neural systems contributing towards its control The ability to focus attention on a single task or relevant stimulus with minimal distraction by extraneous stimuli is an important skill for successfully performing most cognitive tasks. Attention that is directed by voluntary intention on the part of an individual to a particular cue is called top-down or goal-directed attention, while attentional focus that is temporarily controlled by events in the environment, is referred to as bottom-up or stimulus driven attentional control (Corbetta & Shulman, 2002; Fox, Corbetta, Snyder, Vincent, & Raichle, 2006) Studies examining activation of brain areas using fMRI while individuals perform various attentional tasks, have characterized the specific brain regions involved in voluntary vs. stimulus driven modes of attention. Top-down voluntary attention is controlled by the dorsal frontoparietal network that consists of parts of the intraparietal sulcus and frontal eye fields (FEF) bilaterally, whereas the bottom-up stimulus driven network is right lateralized and includes the ventral prefrontal cortex (VLPFC) and the temporoparietal junction (TPJ). The TPJ is believed to act as a circuit-breaker in the 23 process of disengaging attention (Corbetta & Shulman, 2002). Attentional orienting in visual space is defined as allocating attention to a sensory stimulus (Posner, 1980). Orienting can be measured using various paradigms which measure reaction times to spatial or non-spatial cues (e.g. attentional networks test (ANT), spatial cue paradigm. Attentional capture can be exogenous or automatic (e.g., a flash of light in the periphery of vision). These shifts of attention can be influenced by the features of the target stimuli that are shared by the cue or the distractor. The contingent attentional capture hypothesis states that there is an involuntary shift in attention to a given stimulus event if the stimulus and the task involved share the same properties (Folk, Remington, & Johnston, 1992; Folk, Remington, & Wright, 1994; Raz & Buhle, 2006). Orienting attention has been hypothesized to be a function of the interaction of these two systems. The attentional system can be "programmed" to selectively pay attention to a property of a target when the spatial location is unknown. The occurrence of an involuntary shift of attention will depend on the similarities of the properties exhibited by the target and the cue (Folk, Remington, & Johnston, 1992). The interaction between the dorsal and ventral systems of attention can be assessed through a variety of tasks, including the contingent attentional capture paradigm (Folk, Leber, & Egeth, 2002; Folk, Remington, & Johnston, 1992; Serences et aI., 2005). This task was originally devised to investigate the neural basis of the interaction between voluntary attentional control and stimulus-driven attentional capture. Testing the performance of individuals on the contingent capture task reveals their ability to modulate attentional focus, and ignore distracting stimuli. Thus this test can be used to 24 measure the control of voluntary attention by measuring the impact of attention on post­ perceptual processes like response selection. Evidence for the trainability ofattentional skills Previous research has shown that attention is a flexible and trainable skill. For example computerized training involving attention tasks (e.g. sustained attention, selective attention, orienting of attention and executive attention) have shown improvements in non-trained measures of attentional function like non-verbal complex reasoning tasks in typically developing children and adults and also in children and young adults with attention deficit hyperactivity disorder (ADHD). Green & Bavelier found that individuals playing computer action games demonstrated better attentional abilities at central and peripheral locations both as compared to the non-garners. La Pera et al. used somatosensory oddball paradigm in elderly individuals to demonstrate trainability of voluntary oriented attention and not automatic attention (Klingberg et aI., 2005; Klingberg, Forssberg, & Westerberg, 2002a, 2002b; Rueda, Posner, & Rothbart, 2005; Rueda, Rothbart, McCandliss, Saccomanno, & Posner, 2005; Shalev, Tsal, & Mevorach, 2007) (Green & Bavelier, 2006; Le Pera, Ranghi, De Armas, Valeriani, & Giaquinto, 2005). In addition to the use of computerized games and programs for attentional training, many clinics and hospitals are using meditation or mindfulness training to improve attentional focus and self-regulation of mental events (Cahn & Polich, 2006; Kabat-Zinn et aI., 1998; Linda, Zenovia, Eileen, Maureen, & Michael, 2001). As defined by Shapiro (1982), meditation is a family of techniques which have in common a 25 conscious attempt to focus attention in a non-analytical way, and an attempt not to dwell on discursive, ruminating thought. Discursive and ruminating thoughts imply digressive ideas which depart from the main focus of attention. In concentrative meditation techniques, attentional focus is maintained on a static or dynamic object (e.g. the breath, a syllable or a visual object) by exerting voluntary control. Hence an element that contributes to improved meditational focus involves attentional training (Dietrich, 2003; Lutz, Greischar, Rawlings, Ricard, & Davidson, 2004) (Chan & Woollacott, 2007). Superior abilities to exert voluntary control of attention in meditation practitioners as compared to non-practitioners have been documented in the above mentioned studies. These abilities have been studied both as trait effects (Le., long-term changes in attentional function) and also as state effects (Le., transient changes resulting from a single session of meditation). Various studies have assessed the attentional abilities of meditators behaviorally (Chan & Woollacott, 2007; Jha et al. 2007; Slaghter et aI, 2007). One cross-sectional study comparing meditators to age, gender and education-matched controls showed that executive attention (as measured by reduced interference effects on the Stroop task) showed a trait effect of improved function in meditators, and this was correlated with minutes of meditation practice per day (Chan & Woollacott, 2007). A second study examined the ability of meditators to process two temporally-close stimuli, which compete for limited attentional resources, in an 'attentional-blink" experiment. They found a smaller attentional blink after three months of intensive meditation training indicating that meditation training can cause increased control over the use of limited attentional processes (Slaghter et aI, 2007). 26 Though these studies have focused on measuring and defining differences in the attentional abilities of meditation practitioners and non-practitioners, the interpretability of their results for meditation improving attention is difficult due to methodological limitations. For example, self-selection biases and between-group differences in either genetic or environmental factors are possible confounds in correlation studies. In addition most of the longitudinal studies have not used appropriate control training groups with equivalent lengths of training periods on a different type of task, but have used wait-listed control subjects. In this situation, placebo effects (the simple process of administering any treatment) are not taken into account and it is not clear to what extent the particular activity contributes to improvements in attention (Davidson et aI., 2003; Lehrer, Woolfolk, Rooney, McCann, & Carrington, 1983). Thus, there is a need for carefully controlled, randomized, clinical trials on the effects of meditation training on attentional processing in individuals. Finally, previous research characterizing the effects ofmeditation on mental states of individuals have proposed meditation training as containing both elements of relaxation and attentional focus, being defined as a relaxed yet heightened alert state (Dietrich, 2003). It would be helpful to separate out the contributions of these two aspects ofmeditation training to assess improvements in attentional function. Meditation and Contingent Capture ofattention In order to address these issues, the current study used the contingent attentional capture paradigm (Folk et a1.1992) to determine if eight weeks of meditation training (40 min/day, 5 days/wk) would influence voluntary top-down attentional control (the dorsal 27 attentional network) and concomitant abilities to ignore distracting stimuli (the ventral attentional network). The study was designed as a randomized controlled trial, with the control group being trained in progressive muscle relaxation. Meditation training focused attention on the breath whereas in progressive muscle relaxation training attention shifted from one muscle group to another in order to relax the various muscle groups of the body. The study aimed to determine whether any attentional improvements could be specifically attributed to meditation training, rather than a general training placebo effect or relaxation. If an improvement in the cost of invalid cues, after training, was seen only in the meditation group as compared to the relaxation group then according to the attention training hypothesis, meditation training would be associated with enhancing the abilities of the dorsal attention system to overcome reflexive orienting. In contrast, if the same cost of invalid cues was seen in both groups, post-training, then the effect of training would be seen as an outcome of relaxation components in either the trainings or a placebo effect. We hypothesized that meditation training would improve performance on the contingent attentional capture paradigm more than relaxation training, due to the attentional focus involved in meditation. In order to investigate attentional capture we replicated Experiment 1 and 2 from Folk et al (1992). In this task, subjects are asked to classify a target character, which can occur at one of four locations around the fixation point, as either an 'x' or an '='. On separate blocks of trials subjects are instructed to either respond to an 'abrupt onset' where the target appears with no other characters or to a red target among three other white distractors. In addition, across blocks subjects were provided with different spatial 28 cues that preceded the onset of the target: either they were given no cue, a central cue (which provides no information about the location of the target, a valid location cue, or an invalid location cue. On abrupt onset blocks, subjects are not surprisingly faster at identifying validly cued targets than no cue targets. Subjects are slower, however, to respond to abrupt onset targets when they are invalidly cued than when there is no cue, suggesting that the invalid cues have "captured" their attention, despite the fact that subjects know they are irrelevant (since these trials are blocked, the subjects know the invalid cues are not informative). In the abrupt onset condition when the cue is invalid it produces a cost on reaction time as it shares the onset property with the target. However in the color target condition, there is no cost of the invalid cue as the subject identifies the target based on the color. With attentional training we would expect that the cost in reaction times for invalid times should be lesser or the benefits of the valid cues should be greater. We expect to see a smaller cost for invalid cues after training in the meditation group as compared to the relaxation group. Methods Subjects This was a pretraining- posttraining research design. Each participant was tested before and after training under similar testing conditions (same time of day and in the same testing room). Participants (n=30) were recruited from an undergraduate Alternative and Complementary Medicine class. Based on the information provided in a lifestyle questionnaire, the participants were assessed on age, gender, physical activity, sleep, education level and GPA and then pseudo-randomly assigned to either the meditation or 29 relaxation training group. Subject demographics are shown in Table I. Physical activity scores were computed based on occupational demands and the exercise regimen followed by an individual. Participants with a history of head injury, concussion or learning disabilities were excluded. Participants with previous experience in meditation (n=2) participated in class activity for the extra-credit opportunity; however their data were not incorporated in analysis. The participants practiced the assigned activity (meditation/relaxation) for 20 minutes in class and 20 minutes outside class (a total of 40 min/day) for at least 5 days a week. This training regimen was maintained for eight weeks and their compliance with the practice was measured with weekly logs. Tablel. Demographics of the participants in the study Sr.No Meditation Relaxation 1. Age (Mean) 23.lyears( S.D.3.l) 22.8 years(S.D. 4.8) 2. Gender 12F,3M IIF,4M 3. Physical activity 2.2 2.8 4. Educational level 15.9 15.8 (years) 5. GPA 3.3 3.2 The participants in both groups were given handouts describing the daily practice of meditation or progressive muscle relaxation before training began. In-class sessions were guided by expert instructors. At the beginning of a session, the meditation group 30 was instructed to sit in an erect comfortable posture with the spine elongated. The subjects were instructed to focus on the breath and silently repeat the syllable 'Om' on every in-breath and out-breath. The subjects were then intermittently reminded to let go of any thoughts arising and focus on their breathing. Since the subjects focused on their breathing, we define this meditation training as focused attentional training. The relaxation group was instructed in progressive muscle relaxation, with individuals being instructed to focus on the different muscle groups of the body in succession (starting with muscles of the feet and working upwards toward the face). At the end of the single sequence of relaxation instructions, the participants were instructed to let their mind relax and wander to whatever thoughts arose. None of the participants fell asleep during the relaxation training; however, they reported in weekly logs mind­ wandering to relaxing thoughts. To prevent any expectations about the effectiveness of a particular training for improving attentional abilities, the two groups were given information about meditation and relaxation as an alternative medical therapy for stress reduction in class and led to believe that both the training paradigms were equally effective in training. Any biases in the beliefs of participants about training were assessed with the help of an exit questionnaire at the end of the training program. The participants filled out the KIMS (Kentucky Inventory for Mindfulness Skills) assessment self report after completing the entire training program at the end of the study. The inventory had questions which could assess changes in ability to observe experiences, ability to describe experiences, awareness of experience, and non­ judgmental acceptance of experiences. 31 Procedure The behavioral task, the contingent attentional capture task (Folk, Remington, & Johnston, 1992) is depicted in Figure 1. The subjects were seated about 40 cm away from the screen. Depending upon the goal-set the subjects had two target conditions: abrupt onset condition and color target condition. In the abrupt onset condition, participants were instructed about the cue-target location relationship and a single 'x' or '=' appeared on the target display. The participants responded to the type of target seen. In the color target condition, the information about cue-target location was known to participants but the instruction was to respond to a red character. Thus, there were four cue-target conditions: the valid cue (target appears in the same spatial location as the precue), invalid cue (target does not appear in the same spatial location as the precue), center cue (the cue was always in the center), and the no cue condition. In each block of trials, the cue condition remained the same and the subjects were informed about the cue condition in advance. Thus subjects were completely aware of the spatial relationship between the cue and the target. (The terms valid and invalid are often used in reference to expectancy information that is given by a cue. In this paradigm we use exogenous cues, and the terms refer only to the relationship between the positions in space of the cue and target.) The subjects were asked to respond to the target character they saw on the display (a left key for 'X' and a right key for '='). Reaction times and accuracy were measured for each trial. 32 TARGET DISPlAY FlXAllCN DISPlAY CUE DISPl.AV 10m. 6Om. 100m1D(Xl~ 140Ct m.. 0 IE]0 0 00 0 ___.JI CNSerrTAAGET,~O 0 0 0 0 0 ~ 0 ~ l\1DLJ .­ I.§ Figure 1. Task display for contingent capture paradigm (from Folk and Remington 1992) Stimuli were presented on a 36x29cm CRT monitor at 800x600 resolutions with a viewing distance of approximately 40cm. The E-Prime computer program was used to display stimulus sequences and to collect subject responses from a keyboard. Each stimulus sequence consisted of a fixation screen displayed for 1000-1400 ms, a 50 ms cue, a second fixation screen displayed for 100 ms, and finally the target display for 50ms. Although the display disappeared immediately, the next trial was initiated only after the subject generated a response to the previous trial. The fixation display consisted of five squares, each with a side length measuring 1.2 deg visual angle. The squares were arranged as a '+', with a square on the top, bottom, left, right, and center. The four outer squares were each located at a visual angle of 5.1 deg from the center. The boxes were light grey in color, against a black background. The cue display consisted of four small white circles centered approximately .3 deg from the sides of one of the five boxes. Cue circles subtended a visual angle of .36deg. The target display 33 consisted of the original fixation display, with either the symbol 'X' or '=' appearing inside one of the squares. The target symbols subtended a visual angle of .5 deg and were colored white. Design This was a pre-training- post-training research design. The dependent variables measured were reaction times and accuracy. Reaction times, measured in milliseconds, are a quantitative variable, whereas accuracy, a qualitative variable, was coded numerically, i.e. 1 for accurate responses and 0 for inaccurate responses. Within subject independent variables were time, cue condition, and target conditions, which are all qualitative variables. Time refers to testing time and had two levels, pre-training and post-training. Cue condition had four levels depending on the spatial location of cues, no cue, center cue, invalid cue, and valid cue. Depending on the shape of the target there were two target conditions, 'x' target and '=' target. However, these two target conditions did not bring about any statistically significant impact so are reported together. The between subjects factor was the training group which had two levels: meditation group and relaxation group. Using the above mentioned design we assessed the impact of training on reaction times and accuracy under various cue and target conditions across both the training groups. ------------------------- 34 Results and Discussion Accuracy analysis The overall accuracy pattern on pre-training testing was similar to the original study (Folk, 1992). There was no significant change in accuracy as a function of training. The results are summarized in Table 2. Table 2. Accuracy as a function of Target-Property condition and Training Onset Accuracy No Cue Valid Invalid Center Meditation Pre 0.93 0.94 0.96 0.95 Post 0.93 0.95 0.95 0.96 Relaxation Pre 0.94 0.92 0.92 0.92 Post 0.96 0.96 0.92 0.92 Color Accuracy No Cue Valid Invalid Center Meditation Pre 0.91 0.93 0.92 0.92 Post 0.93 0.93 0.90 0.93 Relaxation Pre 0.91 0.93 0.92 0.91 Post 0.92 0.85 0.89 0.88 No significant interaction effect of training (time) or type of training (group) was seen (F (1, 23) =.498). No significant effect of cue or condition was found on accuracy and there was no significant cost of invalid cues on accuracy or a benefit of valid cues on accuracy. However, we explored for a possibility of a speed- accuracy trade off in the next section of reaction time analysis. 35 Reaction Time Analysis Global effect on speed We will first discuss the effects of the abrupt onset condition. In this condition there is a contingent capture of attention as the target property (abrupt onset) and task-set is the same. Table 3 shows the mean of reaction times pre- and post-training for both meditation and relaxation groups for each of the cue conditions in the contingent attentional capture task. Table 3. Reaction times (in ms) as a function of training and cue conditions OnsetRT No Cue Valid Invalid Center Meditation Pre 408.8 397.3 489.9 432.2 Post 396.1 391.6 463.8 431.0 Relaxation Pre 427.3 419.1 478.2 460.6 Post 412.5 408.2 455.5 428.9 ColorRT No Cue Valid Invalid Center Meditation Pre 477.29 454.32 483.80 466.06 Post 459.12 435.38 478.26 452.61 Relaxation Pre 473.11 475.52 486.35 488.96 Post 460.02 444.42 482.95 452.97 Reaction times beyond +/- 3 Std. Dev with respect to the mean of each subject's mean reaction times were excluded from analysis. No significant interaction effect of time*group was found for any of the four cue conditions: no cue condition (F(1, 23)<1,), valid cue condition (F(1,23).05, p=.306). Thus, an overall speeding or slowing of reaction times was not seen ruling out a practice effect. From Table 2 and Table 3 we see 36 that in the valid cue condition the groups were faster but their accuracy was lower, whereas in the invalid cue condition or no cue conditions the accuracy was greater but reaction times were slower. This suggests the subjects were conservative in making their responses. Effect ofmisdirected or invalid cues on task performance The cost of an invalid cue (the slowing of reaction times when the cue is invalid) is the difference between the reaction times required for the invalid cue condition and the center cue condition. We used the center cue condition as a control condition as it does not help in orienting or conflict resolution but serves the purpose of alerting the subject, unlike a no cue condition. The costs of invalid cues are shown in Table 4. Table 4. Cost of invalid cue as a function of target-property condition Onset Color Meditation Pre 57.7 17.7 Post 32.7 25.7 Relaxation Pre 17.6 -2.6 Post 26.6 30.0 The cost of an invalid cue is greater for the onset target condition as compared to the color target condition, which is similar to what would be expected from the contingent capture hypothesis. Figure 2 depicts the costs of invalid cues and the benefit of valid cue as a function of training and group. We will discuss the onset target condition in the following section. 37 Onset-target condition: 700 1 Cil 60.0 j .§. 500 OJ .S 40.0 0 Benefit ::::l U • Cost'0 30.0 1 1 ~ 20.0 I ffi 10.0 0.0 pre pre post Meditation Relaxation Figure 2. The effect of cuing as a function of training and group Change in costs ofinvalid cues On collapsing all the subjects into one group, no significant mam effect of training was seen (F (1, 23) <1). We tried to assess the impact of training group on the cost of invalid cues. We did not see significant differences across the groups (F (1, 23) = 2.42, p=.133). However, from figure 2, we find a trend towards reduction in cost of the invalid cue in the meditation group and no such trend in the relaxation group. These results are to be interpreted with caution as we see a difference in the cost of invalid cues across the two groups before training (Time 1). The two groups were carefully matched based on their lifestyles and hence almost identical costs and benefits of cue conditions were expected at the beginning of the training. However, surprisingly, the meditation group showed a larger cost of invalid cues as compared to the relaxation group. 38 It is possible that since the cost of invalid cues in the relaxation group was small there was no opportunity to observe improvement even if one had occurred. Considering medians of the cost of invalid cues of the respective groups, we selected a subset of trials to rule out the differences. The higher end of the relaxation group and the lower end of the meditation group were compared with relation to their own costs with paired samples T-test. The meditation group (lower than median) did not show any significant change (t (5) =.104 p=.92); neither did the upper half whereas the higher than median relaxation group did show a significant change in the cost of invalid cues (t (1)=44.49 p=.014). The possibility that there was no room for change, due to the low cost of the invalid cue at time 1, cannot be ruled out. Hence we need to interpret these results with caution. Effects o/valid (positive) cuing We assessed the change in benefits as a function of training with repeated measures ANOVA. There was no significant impact of training alone when all the subjects were collapsed into one group (F(1,23)& 'A. U'I.. " " ", I • lO~ .,.,,~ ,'r H 10' l"} '0' la;lJ 1111<:\' 10U'Im '" til II' ::, ... l';9III lIS '" '" '" ,., I!: III in ,n1<. 'I' 13:- !l; 11 '" .t·:c '" lH. ". 1":; If~ 2IG Figure 4. Parietal (right) electrode cluster 60 I" .,'1 .11 'I ,,' ,. ,. , .. I. 'II , III ... .., ,. IrfJ , I Ii/Iff ' ,i. ~, II Ill' . ; , " " " i' lJ "•. '" ,. " IllJ >h II. I/l 1\1 t,t I, It. J•• 'I II Ijl 'r II HI "' Jlf. 1J! 11\': l,·j; il " II: I·" ., ", "> " I;;' ", ,. l'SJ'. '" .!, 11..: Figure 5. Frontal electrode cluster EEG data analysis The dependant variable used in the EEG analysis is amplitude of the wavelets. The independent variables manipulated were time (pre-training and post-training), group (meditation, relaxation), region (Frontal, Occipital, and Parietal) and laterality (left, right and midline). Band or frequency of the wavelets had four levels (delta, theta, alpha and beta.). The task condition under consideration in the present paper has two levels: on task during on-focus blocks and average rest period, The analysis for two conditions, meditative state and rest, were computed to measure the impact of training on the amplitude of wavelets across scalp electrode groups and across frequency bands in 7 second epochs ofEEG without a button press. The 61 amplitudes during the meditative condition were normalized with respect to the averages of the resting state amplitudes for each frequency and region during the pre-training test session. For meditative segments a repeated measures ANOYA was computed to assess the impact of training (pre- training, post-training), frequencies (delta, theta, alpha, beta) in three regions (frontal, occipital and parietal) considering laterality (left, right, midline and entire) across both the groups (meditation and relaxation) on the amplitude of the wavelets. A significant interaction was seen for time*band* region*group (F (6, 96) = 2.32, p=.046). A significant interaction was also seen for band*time*region*laterality (F (18,288) = 3.194 p=.OO). Figure 6a. Wavelets from frontal electrode cluster in focused attention state, prior to training. 62 Figure 6b. Wavelets from frontal electrode cluster In focused attention state, after training. Region-wise analysis To determine changes that occurred in each region, as a follow-up analysis, we computed a repeated measure ANOYA for each region. In the occipital and parietal electrode cluster, a significant interaction time*frequency*laterality was seen (occipital ­ F (9, 144) = 3.84, p=.021; parietal- F (9, 144) = 3.86, p=.02). Also a significant interaction between frequency* group was found (occipital- F (3, 48) = 3.08, p=.05; parietal- F (3, 48) = 3.62 p=.032). These results imply a significant effect of training on frequency and laterality. To further understand the impact of time we conducted a follow­ up analysis. Time-wise analysis A paired t-test (pre training- post training) was computed to measure the impact of training across the two groups. A significant difference was seen in theta (pre-post) and alpha (pre-post) frequency in the meditation group but not in the relaxation group, as 63 you see in Figures 7 and 8. Significant differences are noted in the Table 7. Note that alpha frequencies showed training associated increases in amplitudes for both occipital and right parietal regions, while theta frequencies showed similar increases for only occipital regions. Table 7. Change in alpha and theta amplitudes as a function of time Theta Meditation Occipital t(8)=-2.53 p=.03 Occipital RT t(8)=-3.08 p=.015 Occipital Mid t(8)= -2.59 p=.03 Relaxation t(8)= -1.69 p=.129 t(8)= -2.07 p=.07 t(8)= -1.94 p=.088 Alpha Occipital t(8)= -2.675 p= .028 Occipital RT t(8)=-2.87 p=.021 Occipital Mid t(8)=-2.73 p=.026 Parietal RT t(8)=-2.92 p=.019 t(8)=-1.43 p=.189 t(8)= -1.6 p=.138 t(8)= -2.00 p=.08 t(8)=-1.835 p=.104 64 Effect of Training on Theta frequency In Occipital electrode clulter 1.5 D. .. ~ ~ 0.5 I­ "0 .. .. N E g 0 -0.5 -1 Occipital clustor Figure 7. Effect of meditation ys. relaxation training on normalized theta frequency amplitude for the occipital electrode cluster. The column from left to right include the entire occipital cluster, the midline occipital cluster and right occipital cluster. 65 Effect of Training on Alpha frequency z o z ~ 1 !!!. ~. 0. .. U "'".. .. 3 05 "!1. ~ 0.