UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Functional neuroanatomy of self-generated thought : investigating general brain recruitment, specific… Fox, Kieran C. R. 2016

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
24-ubc_2016_september_fox_kieran.pdf [ 20.88MB ]
Metadata
JSON: 24-1.0308694.json
JSON-LD: 24-1.0308694-ld.json
RDF/XML (Pretty): 24-1.0308694-rdf.xml
RDF/JSON: 24-1.0308694-rdf.json
Turtle: 24-1.0308694-turtle.txt
N-Triples: 24-1.0308694-rdf-ntriples.txt
Original Record: 24-1.0308694-source.json
Full Text
24-1.0308694-fulltext.txt
Citation
24-1.0308694.ris

Full Text

FUNCTIONAL NEUROANATOMY OF SELF-GENERATED THOUGHT: INVESTIGATING GENERAL BRAIN RECRUITMENT, SPECIFIC NEURAL CORRELATES, AND NEURAL ORIGINS USING FUNCTIONAL AND MORPHOMETRIC MAGNETIC RESONANCE IMAGING AND DIFFUSION TENSOR IMAGING  by  KIERAN C. R. FOX  B.A., McGill University, 2008 M.A., The University of British Columbia, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY  in  The Faculty of Graduate and Postdoctoral Studies  (Psychology)   THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  August 2016  © Kieran C.R. Fox, 2016 	   ii	  ABSTRACT	   The human brain is not limited to the here and now, but can roam freely away from the present moment. It has evolved the remarkable capacity to conjure up the past, project itself into hypothetical futures, and envision entirely imaginary worlds and experiences. All of these capacities can be understood as variations on the theme of the brain’s ability to self-generate thoughts and imagery without any direct relation or relevance to the surrounding environment and its continuous sensory inputs. Here we examine the neural basis of such self-generated thought from a variety of angles, aiming to address numerous fundamental questions, including: general brain recruitment associated with self-generated thought; the brain basis of self-generated thought throughout the sleep cycle; specific neural correlates associated with specific types of thought; whether experimental contexts (such as the MRI scanner) influence thought content; whether specific individual patterns or ‘styles’ of thinking are associated with individual neuroanatomical heterogeneities in grey and white matter; which brain regions are necessary and sufficient for the experience of self-generated thought in wakefulness and sleep; and finally, the neuroanatomical origin sites that give rise to self-generated mental content. Substantial and meaningful (if preliminary) answers are provided to each of these questions based on six studies involving a variety of meta-analytic and empirical methods, and a combination of behavioral, functional neuroimaging, and morphometric neuroimaging data.   	   iii	  PREFACE	  	   I prepared the content of this dissertation. The first several sections rely heavily on previously published material on which I was the lead (i.e., first) author and main contributor to the work. Specifically, the introductory section (Chapter 1) on self-generated thought content relies heavily on a published review of thought content in waking and dreaming (Fox et al., 2013, Front. Hum. Neurosci., 7(412), 1-18),  as well as further work detailing the affective content of self-generated thought (Fox et al., 2014, Front. Psychol., 5(1427), 1-4). Chapter 2, a meta-analytic assessment of general brain recruitment associated with self-generated thought, relies heavily on a previously published quantitative meta-analysis of functional neuroimaging studies (Fox et al., 2015, NeuroImage, 111, 611-621). Chapter 3 builds on a previously published meta-analytic assessment (Fox et al., 2013, Front. Hum. Neurosci., 7(412), 1-18) and review (Domhoff and Fox, 2015, Conscious. Cogn., 33, 342-353) of brain recruitment during rapid eye movement (REM) sleep, as well as unpublished work in progress on the neural correlates of the other sleep stages. Chapters 4, 5, and 6 are based on newly-acquired and unpublished data collected by me and on which projects I was the lead researcher. Chapters 7 and 8 rely heavily on an under revision paper synthesizing lesion and single-neuron recording studies of self-generated thought (Fox et al., under revision).    	   iv	  TABLE	  OF	  CONTENTS	  ABSTRACT	  ..............................................................................................................................................	  ii	  PREFACE	  .................................................................................................................................................	  iii	  TABLE	  OF	  CONTENTS	  .........................................................................................................................	  iv	  LIST	  OF	  TABLES	  .................................................................................................................................	  viii	  LIST	  OF	  FIGURES	  ..................................................................................................................................	  ix	  ACKNOWLEDGMENTS	  ..........................................................................................................................	  x	  CHAPTER	  1	  –	  HISTORY	  AND	  OVERVIEW	  OF	  SELF-­‐GENERATED	  THOUGHT	  ......................	  1	  Why	  study	  the	  self-­‐generated	  thought	  of	  the	  ‘resting’	  brain?	  ..........................................................	  5	  The	  ubiquity	  of	  self-­‐generated	  thought	  .................................................................................................................	  6	  Self-­‐knowledge,	  self-­‐improvement,	  and	  mental	  health	  ..................................................................................	  6	  Creativity,	  insight,	  and	  self-­‐generated	  thought	  ..................................................................................................	  8	  Pure	  scientific	  inquiry	  and	  self-­‐generated	  thought	  ..........................................................................................	  9	  The	  content	  of	  self-­‐generated	  thought	  ..................................................................................................	  10	  The	  valence	  of	  self-­‐generated	  thought	  ..................................................................................................	  13	  Cognitive	  control	  and	  metacognition	  in	  self-­‐generated	  thought	  .................................................	  15	  CHAPTER	  2	  –	  STUDY	  1:	  GENERAL	  NEURAL	  CORRELATES	  OF	  SELF-­‐GENERATED	  THOUGHT:	  A	  META-­‐ANALYTIC	  ASSESSMENT	  OF	  FUNCTIONAL	  NEUROIMAGING	  INVESTIGATIONS	  ................................................................................................................................	  16	  Introduction	  ...................................................................................................................................................	  16	  Methods	  ...........................................................................................................................................................	  18	  Search	  strategy	  ..............................................................................................................................................................	  18	  Study	  inclusion	  and	  exclusion	  criteria	  ................................................................................................................	  19	  Adjustment	  of	  peak	  coordinates	  ............................................................................................................................	  22	  Activation	  likelihood	  estimation	  (ALE)	  meta-­‐analysis	  ................................................................................	  22	  Results	  ..............................................................................................................................................................	  25	  Discussion	  .......................................................................................................................................................	  28	  What	  constitutes	  the	  default	  mode	  network?	  ..................................................................................................	  28	  The	  default	  mode	  network	  and	  self-­‐generated	  thought	  ..............................................................................	  29	  Frontoparietal	  control	  network	  involvement	  in	  self-­‐generated	  thought	  ............................................	  30	  Coupling	  of	  the	  default	  mode	  and	  frontoparietal	  control	  networks	  ......................................................	  31	  Other	  cortical	  areas	  recruited	  by	  self-­‐generated	  thought	  ..........................................................................	  32	  Subcortical	  contributions	  to	  self-­‐generated	  thought?	  ..................................................................................	  34	  Implications	  for	  clinical	  disorders	  of	  self-­‐generated	  thought	  ...................................................................	  35	  Limitations	  of	  the	  meta-­‐analysis	  ............................................................................................................................	  37	  Conclusions	  and	  future	  directions	  ........................................................................................................................	  38	  CHAPTER	  3	  –	  STUDY	  2:	  NEURAL	  CORRELATES	  OF	  SELF-­‐GENERATED	  THOUGHT	  IN	  SLEEP	  AND	  DREAMING:	  A	  PRELIMINARY	  META-­‐ANALYTIC	  ASSESSMENT	  OF	  FUNCTIONAL	  NEUROIMAGING	  INVESTIGATIONS	  ....................................................................	  39	  Introduction:	  the	  multiplicity	  of	  sleep	  ..................................................................................................	  39	  Sleep	  mentation	  as	  self-­‐generated	  thought	  ......................................................................................................	  41	  Sleep	  can	  be	  meaningfully	  dissociated	  into	  stages	  ........................................................................................	  42	  Phenomenological	  dissociation	  of	  sleep	  stages	  ...............................................................................................	  44	  Electrophysiological	  dissociation	  of	  sleep	  stages	  ...........................................................................................	  45	  	   v	  Neurochemical	  dissociation	  of	  sleep	  stages	  .....................................................................................................	  48	  Functional	  neuroimaging	  of	  sleep	  and	  dreaming:	  a	  fourth	  line	  of	  evidence	  ......................................	  50	  Aims	  of	  the	  present	  meta-­‐analysis	  ........................................................................................................................	  51	  Methods	  ...........................................................................................................................................................	  52	  Study	  selection	  ...............................................................................................................................................................	  52	  Activation	  likelihood	  estimation	  (ALE)	  meta-­‐analysis	  ................................................................................	  54	  Reporting	  of	  meta-­‐analytic	  results	  .......................................................................................................................	  55	  Results	  ..............................................................................................................................................................	  55	  Neural	  correlates	  of	  NREM1	  sleep	  ........................................................................................................................	  55	  Neural	  correlates	  of	  NREM2	  sleep	  ........................................................................................................................	  56	  Neural	  correlates	  of	  NREM3/4	  sleep	  (SWS)	  .....................................................................................................	  58	  Neural	  correlates	  of	  REM	  sleep	  ..............................................................................................................................	  60	  Discussion	  .......................................................................................................................................................	  64	  Default	  and	  visual	  network	  activation	  and	  their	  relationship	  to	  self-­‐generated	  thought	  and	  imagery	  throughout	  the	  sleep	  cycle	  .....................................................................................................................	  64	  Brain	  deactivation,	  cognitive	  control,	  and	  meta-­‐awareness	  throughout	  the	  sleep	  cycle	  .............	  67	  Limitations	  ......................................................................................................................................................................	  70	  Conclusions	  .....................................................................................................................................................................	  72	  CHAPTER	  4	  –	  STUDY	  3:	  SPECIFIC	  NEURAL	  RECRUITMENT	  ASSOCIATED	  WITH	  DIFFERENT	  TYPES	  OF	  THOUGHT	  ..................................................................................................	  73	  Introduction	  ...................................................................................................................................................	  73	  Investigating	  abstract	  qualities	  of	  thought	  .......................................................................................................	  73	  Previous	  efforts	  to	  understand	  specific	  neural	  recruitment	  associated	  with	  different	  types	  of	  thought	  ..............................................................................................................................................................................	  76	  Methods	  ...........................................................................................................................................................	  77	  Participants	  .....................................................................................................................................................................	  77	  Experimental	  design	  ...................................................................................................................................................	  78	  Procedure	  and	  fMRI	  thought	  sampling	  paradigm	  ..........................................................................................	  80	  fMRI	  data	  acquisition	  ..................................................................................................................................................	  82	  fMRI	  data	  preprocessing	  and	  analysis	  .................................................................................................................	  83	  Results	  ..............................................................................................................................................................	  85	  Behavioral	  results	  ........................................................................................................................................................	  85	  SIT: stimulus-independent thought.	  .........................................................................................................................	  85	  Orthogonality	  of	  thought	  qualities	  ........................................................................................................................	  85	  Neural	  correlates	  of	  spontaneity	  and	  directedness	  in	  stimulus-­‐independent	  thought	  .................	  86	  Neural	  correlates	  of	  goal-­‐relatedness	  in	  stimulus-­‐independent	  thought	  ............................................	  91	  Neural	  correlates	  of	  affective	  valence	  in	  stimulus-­‐independent	  thought	  ............................................	  93	  Discussion	  .......................................................................................................................................................	  95	  Neural	  correlates	  of	  spontaneously	  unfolding	  and	  intentionally	  directed	  thoughts	  ......................	  95	  Neural	  correlates	  of	  goal-­‐related	  stimulus-­‐independent	  thoughts	  ........................................................	  97	  Neural	  correlates	  of	  positively-­‐valenced	  thoughts	  ........................................................................................	  97	  Limitations	  ......................................................................................................................................................................	  99	  Conclusions	  and	  future	  directions	  .....................................................................................................................	  101	  CHAPTER	  5	  –	  STUDY	  4:	  DOES	  THE	  MRI	  SCANNER	  ENVIRONMENT	  INFLUENCE	  THOUGHT	  CONTENT?	  ....................................................................................................................	  103	  Introduction	  .................................................................................................................................................	  103	  Methods	  .........................................................................................................................................................	  104	  Participants	  ..................................................................................................................................................................	  104	  	   vi	  Statistical	  analyses	  ....................................................................................................................................................	  104	  Results	  ............................................................................................................................................................	  105	  Average	  thought	  content	  across	  contexts	  .......................................................................................................	  105	  Stability	  of	  thought	  content	  patterns	  at	  the	  individual	  level	  ..................................................................	  107	  Discussion	  .....................................................................................................................................................	  109	  The	  MRI	  scanner	  environment	  does	  not	  appreciably	  influence	  thought	  content	  .........................	  109	  Individual	  patterns	  of	  thought	  content	  are	  stable	  over	  time	  and	  across	  contexts	  .......................	  111	  CHAPTER	  6	  –	  STUDIES	  5	  AND	  6:	  INVESTIGATING	  THE	  RELATIONSHIP	  BETWEEN	  INDIVIDUAL	  DIFFERENCES	  IN	  SELF-­‐GENERATED	  THOUGHT	  AND	  BRAIN	  STRUCTURE	  USING	  MORPHOMETRIC	  NEUROIMAGING	  ...............................................................................	  112	  Introduction:	  the	  relevance	  of	  interindividual	  neuroanatomical	  heterogeneity	  ................	  112	  Previous	  research	  using	  morphometric	  neuroimaging	  to	  investigate	  self-­‐generated	  thought	  ...........................................................................................................................................................................................	  114	  The	  need	  for	  further	  morphometric	  neuroimaging	  of	  self-­‐generated	  thought	  ..............................	  115	  Methods	  .........................................................................................................................................................	  116	  Participants	  ..................................................................................................................................................................	  116	  Procedure	  and	  behavioral	  thought-­‐sampling	  session	  ...............................................................................	  117	  Calculation	  of	  individual	  difference	  scores	  in	  self-­‐generated	  thought	  ...............................................	  118	  MRI	  data	  acquisition	  ................................................................................................................................................	  118	  Preprocessing	  of	  3D	  T1	  anatomical	  images	  ...................................................................................................	  119	  Preprocessing	  of	  DTI	  images	  ................................................................................................................................	  120	  Voxel-­‐based	  morphometry	  analysis	  of	  grey	  matter	  concentration	  .....................................................	  120	  Tract-­‐based	  analysis	  of	  fractional	  anisotropy	  ...............................................................................................	  121	  Behavioral	  results:	  individual	  differences	  in	  thought	  content	  ...................................................	  123	  Morphometric	  neuroimaging	  results	  I:	  grey	  matter	  concentration	  .........................................	  124	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  intentionally	  directed	  stimulus-­‐independent	  thought	  ...............................................................................................................................................	  124	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  goal-­‐related	  stimulus-­‐independent	  thought	  ...........................................................................................................................................................................	  127	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  positively-­‐valenced	  stimulus-­‐independent	  thought	  ...............................................................................................................................................	  129	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  negatively-­‐valenced	  stimulus-­‐independent	  thought	  ...............................................................................................................................................	  131	  Morphometric	  neuroimaging	  results	  II:	  fractional	  anisotropy	  ..................................................	  132	  Discussion	  .....................................................................................................................................................	  134	  Grey	  matter	  concentration	  and	  intentionally	  directed	  thought	  ............................................................	  134	  Grey	  matter	  concentration	  and	  goal-­‐related	  thought	  ................................................................................	  134	  Grey	  matter	  concentration	  and	  the	  valence	  of	  thought	  ............................................................................	  135	  White	  matter	  integrity	  and	  self-­‐generated	  thought	  tendencies	  ............................................................	  136	  Cellular	  foundations	  of	  macroscale	  morphological	  changes	  ..................................................................	  138	  Limitations	  ...................................................................................................................................................................	  139	  CHAPTER	  7	  –	  IDENTIFYING	  THE	  NECESSARY	  AND	  SUFFICIENT	  NEURAL	  BASIS	  OF	  SELF-­‐GENERATED	  THOUGHT:	  THE	  VALUE	  OF	  NEUROPSYCHOLOGICAL	  LESION	  STUDIES	  ..............................................................................................................................................	  141	  Introduction	  .................................................................................................................................................	  141	  Identifying	  necessary	  and	  sufficient	  neural	  substrates	  of	  self-­‐generated	  thought	  .............	  142	  CHAPTER	  8	  –	  NEURAL	  ORIGINS	  AND	  ONTOGENY	  OF	  SELF-­‐GENERATED	  THOUGHT	  AS	  REVEALED	  BY	  HUMAN	  INTRACRANIAL	  ELECTROPHYSIOLOGY	  ......................................	  148	  	   vii	  Introduction	  .................................................................................................................................................	  148	  Neural	  origins	  of	  self-­‐generated	  thought	  ...........................................................................................	  149	  The	  subsequent	  ontogeny	  of	  self-­‐generated	  thought	  ....................................................................	  154	  CHAPTER	  9	  –	  SUMMARY	  AND	  CONCLUSIONS	  .........................................................................	  159	  Self-­‐generated	  thought	  in	  general	  is	  associated	  with	  widespread	  brain	  recruitment	  .......	  159	  Self-­‐generated	  thought	  and	  imagery	  in	  sleep	  and	  dreaming	  share	  a	  neural	  substrate	  with	  waking	  thought	  ...........................................................................................................................................	  160	  Distinct	  categories	  of	  thought	  can	  be	  resolved	  at	  the	  neural	  level	  with	  fMRI	  .......................	  160	  The	  MRI	  scanner	  environment	  does	  not	  appreciably	  affect	  self-­‐generated	  thought	  content	  ..........................................................................................................................................................................	  161	  Interindividual	  differences	  in	  thought	  content	  are	  associated	  with	  individual	  neuroanatomical	  heterogeneity	  ............................................................................................................	  162	  Lesions	  studies	  point	  to	  necessary	  neural	  substrates	  of	  self-­‐generated	  thought	  ................	  162	  Human	  intracranial	  electrophysiology	  suggests	  a	  medial	  temporal	  lobe	  origin	  site	  for	  self-­‐generated	  thought	  ......................................................................................................................................	  163	  REFERENCES	  .....................................................................................................................................	  164	  	  	   	  	   viii	  LIST	  OF	  TABLES	  	  TABLE	  1.	  MAJOR AREAS OF INQUIRY IN THE COGNITIVE NEUROSCIENCE OF SELF-GENERATED THOUGHT.	  ............................................................................................................................................................................................	  4	  TABLE	  2.	  FUNCTIONAL NEUROIMAGING STUDIES EXAMINING SELF-GENERATED THOUGHT.	  ..........	  24	  TABLE	  3.	  META-ANALYTIC RESULTS: BRAIN AREAS CONSISTENTLY ACTIVATED BY MIND-WANDERING AND RELATED SELF-GENERATED THOUGHT PROCESSES.	  ...............................................	  26	  TABLE	  4.	  FREQUENCY OF SUBJECTIVE REPORTS OF COGNITIVE ACTIVITY AND DREAMING ACROSS SLEEP STAGES.	  ..........................................................................................................................................................	  45	  TABLE	  5.	  NEUROCHEMICAL PROFILES OF THE VARIOUS STAGES OF SLEEP AS COMPARED TO WAKING.	  ............................................................................................................................................................................................	  50	  TABLE	  6.	  BRAIN STATES INVESTIGATED, WITH NUMBER OF CONTRIBUTING STUDIES AND FOCI, AND NUMBER OF META-ANALYTIC RESULTS OBTAINED.	  ..............................................................................	  52	  TABLE	  7.	  STUDIES INCLUDED IN META-ANALYSES OF SLEEP AND DREAMING.	  ...........................................	  53	  TABLE	  8.	  ACTIVATIONS AND DEACTIVATIONS ASSOCIATED WITH NREM1 SLEEP.	  ....................................	  56	  TABLE	  9.	  ACTIVATIONS AND DEACTIVATIONS ASSOCIATED WITH NREM2 SLEEP.	  ....................................	  57	  TABLE	  10.	  ACTIVATIONS AND DEACTIVATIONS ASSOCIATED WITH NREM3/4 SLEEP (SWS).	  ...............	  59	  TABLE	  11.	  ACTIVATIONS AND DEACTIVATIONS ASSOCIATED WITH REM SLEEP.	  .......................................	  61	  TABLE	  12.	  MAJOR QUALITIES OF SELF-GENERATED THOUGHT UNINVESTIGATED AT THE NEURAL LEVEL.	  .................................................................................................................................................................................................	  75	  TABLE	  13.	  BEHAVIORAL DATA FOR ALL FMRI RUNS.	  ......................................................................................................	  85	  TABLE	  14.	  PAIRWISE CORRELATIONS BETWEEN THOUGHT QUALITIES.	  ............................................................	  86	  TABLE	  15.	  ACTIVATION	  PEAKS	  FOR	  SPONTANEOUSLY	  UNFOLDING	  THOUGHTS	  (>	  INTENTIONALLY	  DIRECTED	  THOUGHTS)	  ................................................................................................................................................................	  88	  TABLE	  16.	  ACTIVATION PEAKS FOR GOAL-RELATED THOUGHTS (> GOAL-UNRELATED THOUGHTS)	  ..................................................................................................................................................................................................................	  91	  TABLE	  17.	  ACTIVATION PEAKS FOR POSITIVELY-VALENCED THOUGHTS (> NEUTRAL + NEGATIVE THOUGHTS)	  ......................................................................................................................................................................................	  93	  TABLE	  18.	  PARTICIPANT FEATURES FOR CONTEXT COMPARISONS.	  ...................................................................	  104	  TABLE	  19.	  INDIVIDUAL PATTERNS OF THOUGHT CONTENT	  .....................................................................................	  123	  TABLE	  20.	  GREY MATTER CONCENTRATION PEAKS FOR HIGHER FREQUENCY OF INTENTIONALLY DIRECTED THOUGHTS	  ............................................................................................................................................................	  125	  TABLE	  21.	  GREY MATTER CONCENTRATION PEAKS FOR HIGHER FREQUENCY OF GOAL-RELATED THOUGHTS	  .....................................................................................................................................................................................	  127	  TABLE	  22.	  GREY MATTER CONCENTRATION PEAKS FOR HIGHER FREQUENCY OF POSITIVELY-VALENCED THOUGHTS	  ..........................................................................................................................................................	  129	  TABLE	  23.	  GREY MATTER CONCENTRATION PEAKS FOR HIGHER FREQUENCY OF NEGATIVELY-VALENCED THOUGHTS	  ..........................................................................................................................................................	  131	  TABLE	  24.	  CANDIDATE CELLULAR MECHANISMS PUTATIVELY UNDERLYING MACROSCALE STRUCTURAL DIFFERENCES IN GRAY AND WHITE MATTER	  .......................................................................	  139	  TABLE	  25.	  SUMMARY	  OF	  HUMAN	  ELECTROPHYSIOLOGY	  STUDIES	  DEMONSTRATING	  ELICITATION	  OF	  MEMORIES,	  THOUGHTS,	  OR	  HALLUCINATORY,	  DREAM-­‐LIKE	  EXPERIENCES.	  .................................................	  153	   	  	   ix	  LIST	  OF	  FIGURES	  	  FIGURE 1. AVERAGE AFFECT IN SELF-GENERATED THOUGHT...................................................................14 FIGURE 2. META-ANALYTIC FMRI BRAIN ACTIVATIONS..............................................................................27 FIGURE 3. BRAIN NETWORKS AND META-ANALYTIC ACTIVATIONS........................................................28 FIGURE 4. EEG CORRELATES OF SLEEP…………..............................................................................................47 FIGURE 5. BRAIN ACTIVATIONS AND DEACTIVATIONS ACROSS SLEEP STAGES...................................63 FIGURE 6. MODEL OF BRAIN ACTIVITY ACROSS SLEEP STAGES................................................................69 FIGURE 7. THOUGHT PROBE QUESTIONS AND SEQUENCE...........................................................................81 FIGURE 8. ACTIVATIONS FOR SPONTANEOUSLY UNFOLDING THOUGHTS.............................................88 FIGURE 9. ACTIVATIONS FOR INTENTIONALLY DIRECTED THOUGHTS...................................................89 FIGURE 10. ACTIVATIONS FOR GOAL-RELATED THOUGHTS.......................................................................91 FIGURE 11. ACTIVATIONS FOR POSITIVELY-VALENCED THOUGHTS........................................................93 FIGURE 12. AVERAGE THOUGHT PROPORTIONS...........................................................................................103 FIGURE 13. RADAR PLOT OF THOUGHT PROPORTIONS...............................................................................104 FIGURE 14. STABILITY OF THOUGHT CONTENT PATTERNS ACROSS CONTEXTS.................................105 FIGURE 15. WHITE MATTER TRACTS: REGIONS OF INTEREST...................................................................119 FIGURE 16. GREY MATTER CONCENTRATION AND INTENTIONALLY DIRECTED THOUGHT............123 FIGURE 17. GREY MATTER CONCENTRATION AND GOAL-RELATED THOUGHT..................................125 FIGURE 18. GREY MATTER CONCENTRATION AND POSITIVELY-VALENCED THOUGHT...................127 FIGURE 19. GREY MATTER CONCENTRATION AND NEGATIVELY-VALENCED THOUGHT................128 FIGURE 20. THOUGHT TENDENCIES AND MEAN FRACTIONAL ANISTROPY..........................................130 FIGURE 21. BRAIN LESIONS INFLUENCING SELF-GENERATED THOUGHT.............................................143 FIGURE 22. MEDIAL TEMPORAL LOBE ORIGIONS OF SELF-GENERATED THOUGHT...........................149               	  	   	  	   x	  ACKNOWLEDGMENTS	  	  	   Many	  thanks,	  first	  and	  foremost,	  to	  my	  parents,	  my	  three	  brothers,	  and	  the	  enormous	  extended	  Fox	  family	  throughout	  British	  Columbia	  and	  around	  the	  world	  for	  their	  ongoing	  support,	  hospitality,	  and	  interest	  in	  my	  doctoral	  research.	  Huge	  credit	  is	  due	  to	  my	  supervisor	  for	  six	  years,	  Kalina	  Christoff,	  who	  blessed	  me	  with	  the	  greatest	  gift	  a	  mentor	  can	  give:	  absolute	  freedom	  to	  research	  anything	  I	  wanted.	  This	  freedom	  has	  led	  quite	  directly	  to	  the	  eclectic,	  but	  hopefully	  still	  coherent,	  body	  of	  work	  I	  have	  produced	  over	  the	  past	  six	  years.	  Next,	  much	  gratitude	  to	  my	  PhD	  thesis	  committee,	  Beck	  Todd,	  Alex	  Rauscher,	  and	  Steve	  Heine,	  for	  their	  time,	  energy,	  thoughtful	  criticisms,	  and	  most	  importantly,	  their	  willingness	  to	  actually	  read	  through	  this	  entire	  tome	  –	  as	  tiresome	  reading,	  I	  suspect,	  as	  any	  they’ve	  ever	  undertaken.	  Huge	  shout-­‐outs	  are	  due	  also	  to	  my	  peers,	  in	  my	  lab	  and	  out,	  who	  have	  graciously	  played	  guru	  to	  my	  grasshopper	  and	  passed	  on	  the	  oral	  tradition	  of	  neuroimaging	  study	  design,	  analysis,	  and	  interpretation:	  Matt	  Dixon,	  Mike	  Jarrett,	  and	  Melissa	  Ellamil.	  Many	  other	  collaborators	  near	  and	  far	  have	  forged	  my	  thinking,	  assessed	  my	  	  analyses,	  ripped	  up	  my	  writing,	  and	  rendered	  various	  other	  services	  less	  easily	  shoehorned	  into	  amusing	  alliterative(ish)	  phrases,	  all	  to	  my	  betterment	  –	  especial	  mention	  should	  be	  made,	  in	  no	  particular	  order,	  of	  Jess	  Andrews-­‐Hanna,	  Bill	  Domhoff,	  Nathan	  Spreng,	  Peter	  Sedlmeier,	  Tore	  Nielsen,	  Evan	  Thompson,	  Michael	  Liftshitz,	  Sam	  Rumak,	  James	  Floman,	  Nick	  Fitz,	  Liza	  Solomonova,	  Peter	  Reiner,	  and	  Zac	  Irving.	  	   Others	  are	  less	  collaborators	  than	  saviors:	  namely,	  the	  many	  technicians,	  administrators,	  and	  other	  staff	  at	  the	  UBC	  MRI	  Research	  Centre,	  who	  have	  made	  it	  about	  as	  easy	  as	  could	  reasonably	  be	  expected	  for	  me	  to	  peer	  into	  the	  structure	  and	  function	  of	  	   xi	  living,	  oxygen-­‐consuming,	  paramagnetic-­‐hemoglobin-­‐containing,	  human	  brains.	  Eternal	  gratitude	  to	  Trudy	  Harris,	  Linda	  James,	  Alex	  Mazur,	  Linda	  Chandler,	  and	  Laura	  Barlow.	  It	  has	  recently	  been	  found	  that	  our	  lab	  group	  does	  research	  that	  is	  significantly	  more	  ‘cool’	  than	  most	  other	  labs	  (p	  <	  .05).	  Our	  lab	  therefore	  seems	  to	  consistently	  attract	  the	  best	  and	  brightest	  undergraduate	  research	  assistants,	  who	  appear	  to	  be	  ignoring	  the	  heated	  debate	  surrounding	  p-­‐values	  and	  significance	  testing.	  They	  may	  feel	  it	  a	  privilege	  to	  work	  with	  us,	  but	  in	  reality	  it	  is	  we	  who	  are	  lucky	  to	  have	  them.	  Special	  mention	  should	  be	  made	  of	  the	  enthusiasm,	  meticulousness,	  and	  acumen	  of	  Savannah	  Nijeboer,	  with	  whom	  I	  worked	  closely	  for	  years,	  as	  well	  as	  Mara	  L.	  Puertolas	  Lopez,	  Manesh	  Girn,	  and	  Cameron	  C.	  Parro.	  If	  I	  weren’t	  so	  patently	  less	  mature	  than	  all	  of	  these	  outstanding	  young	  researchers,	  I	  would	  say	  something	  along	  the	  lines	  of	  ‘They	  have	  kept	  me	  young	  at	  heart;”	  instead,	  let	  it	  suffice	  to	  say	  that	  I	  have	  greatly	  appreciated,	  and	  benefitted	  from,	  the	  experience	  of	  working	  alongside	  them.	  Penultimately,	  let	  me	  thank	  the	  many	  friends,	  friends	  of	  friends,	  and	  total	  strangers	  who	  hopped	  in	  the	  MRI	  scanner	  in	  the	  name	  of	  science	  and	  advancing	  my	  CV,	  and	  of	  course	  to	  get	  pictures	  of	  their	  own	  brains.	  I	  couldn’t	  have	  done	  it	  without	  the	  participants.	  Also	  kudos	  to	  my	  roommates	  for	  putting	  up	  with	  me	  as	  I	  wrote	  this	  dissertation	  accompanied	  by	  incessant	  electronic	  music	  at	  totally	  unreasonable	  volumes	  and	  times	  of	  day.	  Last,	  but	  by	  no	  means	  least,	  let	  us	  never	  forget	  who	  pays	  the	  bills.	  Many	  thanks	  to	  the	  Mind	  and	  Life	  Institute	  for	  providing	  me	  with	  a	  Francisco	  J.	  Varela	  Research	  Award,	  which	  generously	  funded	  the	  majority	  of	  the	  MRI	  scanning	  costs	  associated	  with	  the	  work	  presented	  in	  this	  dissertation;	  and	  to	  the	  Natural	  Sciences	  and	  Engineering	  Research	  Council	  (NSERC)	  of	  Canada,	  for	  providing	  me	  with	  a	  Vanier	  Canada	  Graduate	  Scholarship	  	   xii	  for	  the	  past	  three	  years	  of	  my	  doctorate,	  as	  well	  as	  awarding	  me	  a	  two-­‐year	  post-­‐doctoral	  fellowship	  which	  will	  allow	  me	  to	  continue	  this	  meandering	  mental	  adventure.	  	  	  	  	  	  	  	  	   	  	   xiii	  	  	  	  	  	  	  	  	  	  γνωθι σεαυτόν	  	  	   1	  CHAPTER	  1	  –	  HISTORY	  AND	  OVERVIEW	  OF	  SELF-­‐GENERATED	  THOUGHT	  	   The human brain is not limited to the here and now, but can roam freely away from the present moment. It has evolved the remarkable capacity to conjure up the past (Squire, 2004), project itself into hypothetical futures (Szpunar et al., 2014), and envision entirely imaginary worlds and experiences (Windt, 2010, Fox et al., 2013, Windt, 2015). All of these capacities can be understood as variations on the theme of stimulus-independence, i.e., the brain’s ability to self-generate thoughts and imagery without any direct relation or relevance to the surrounding environment and its continuous sensory inputs. Given that the least controversial perspective on the evolution of the brain argues that central nervous systems arose to more effectively allow organisms to navigate, manipulate, and anticipate their environments (Striedter, 2005), the ability of the brain to disengage from the surrounding world, and the frequency with which it exercises this capacity, are truly astounding: an estimated 30-50% of our waking hours are accompanied by self-generated and stimulus-independent thought in some form or another (Klinger and Cox, 1987, Kane et al., 2007, Killingsworth and Gilbert, 2010), and a full third of our lives is spent almost completely decoupled from the external world in various stages of sleep – all of which are accompanied by stimulus-independent cognition to some degree (Nielsen, 1999, Hobson et al., 2000, Fox et al., 2013, Domhoff and Fox, 2015). Where do all these thoughts come from? It may be surprising that the seemingly straightforward answers “from the mind,” or “from the brain,” are in fact an incredibly recent, modern understanding of self-generated and creative thought. For nearly all of human history the source of our thoughts, particularly the most important and useful, was almost universally ascribed to divine, demonic, or otherwise external sources of some kind (Jaynes, 1976, Kuijsten, 	   2	  2006, McMahon, 2013). Only in the past few centuries have human beings taken full responsibility for their own mental content, and finally localized thought to the central nervous system – laying the foundations for a protoscience of self-generated thought.  Despite this general acceptance that our own brains are the originators of this self-generated thought and imagery, cognitive neuroscience has, until recently, spent very little time or effort investigating this largely forgotten ‘other half’ of our psychological and neurophysiological functioning (Christoff, 2012). The huge accumulated body of work on task-driven and task-evoked brain function has recently been complemented, however, by a small but rapidly growing number of studies investigating the brain’s ‘intrinsic,’ resting functioning (Raichle, 2010). Aside from immense interest in resting brain metabolism (Raichle and Mintun, 2006) and the role of spontaneous neuronal activity in predictive coding (Clark, 2013), a burgeoning subfield has begun to focus on the self-generated and self-referential cognition and imagery that seem to ubiquitously accompany these ‘default’ modes of brain functioning (Gusnard et al., 2001, Raichle et al., 2001).  These ‘default’ patterns of mental activity have gone by many names, but throughout this work, we will generally use the broad term ‘self-generated thought’ (Andrews-Hanna et al., 2014) rather than the more specific ‘stimulus-independent thought.’ Whereas our own new research contributions (presented in Chapters 4-6) focus specifically on stimulus-independent thought, much of the relevant work in this field has not explicitly distinguished between stimulus-independence and -dependence in the thoughts generated by participants in studies investigating ‘mind-wandering’ (Killingsworth and Gilbert, 2010), ‘daydreaming’ (Singer, 1966), ‘task-unrelated thought’ (Allen et al., 2013), ‘spontaneous thought’ (Dumontheil et al., 2010, Ellamil et al., 2016), ‘undirected thought’ (Christoff, 2012), and ‘mentation’ in sleep and 	   3	  dreaming (Nielsen, 2000). Although much of this thought can be safely assumed to be stimulus-independent, nonetheless stimulus-independence per se is best thought of as a technical term and a quality of thought that must be explicitly assessed during experience sampling. Moreover, self-generated streams of thought can interact dynamically with the environment, for instance in social situations or aesthetic experiences, sometimes making it difficult to determine where environmental influence ends and autonomously-generated thought begins (Dixon et al., 2014). A more vague term such as ‘self-generated thought’ therefore compensates for its ambiguity with its versatility, and allows us to reserve the term stimulus-independent thought for cases in which we are certain its use is justified.   In this introductory chapter we aim to lay the basic groundwork that will support the later, more detailed discussions of new meta-analytic and experimental findings. We first outline a number of reasons for studying the brain’s self-generated thought, followed by brief overviews of the content and affective valence of such thought, as well as the appreciable role played in these largely ‘undirected’ forms of thinking by cognitive control and metacognitive resources. A roadmap is presented in Table 1, which highlights major areas of inquiry in the cognitive neuroscience of self-generated thought, and outlines the 6 studies included in this dissertation that have attempted to provide some preliminary answers to these fascinating and complex problems. A final Chapter 9 provides a summary and general conclusions.      	   4	  Table	  1.	  Major areas of inquiry in the cognitive neuroscience of self-generated thought.	  Chapter (Experiment) Aspect of self-generated thought Central Questions Principal Modality of Investigation Ch. 1 Content, Valence, Cognitive Control What is the basic phenomenological content of self-generated thought, and how is it valenced? What role do cognitive control and metacognition play in these forms of thought? First-person reports; questionnaires Ch. 2 (Study 1) General Brain Recruitment What brain regions and networks are consistently recruited by self-generated forms of thought?  Functional neuroimaging (fMRI and PET) Ch. 3 (Study 2) General Brain Recruitment in Sleep States What brain regions and networks are consistently recruited during the various stages of sleep, which differ in terms of frequency and intensity of self-generated thought and imagery? Functional neuroimaging (fMRI and PET) Ch. 4 (Study 3) Specificity of Brain Recruitment How does brain recruitment vary with the content and type of self-generated thought? Functional neuroimaging (fMRI and PET) Ch. 5 (Study 4) Influence of Context Does the surrounding environment markedly affect thought content? Specifically, does the MRI scanner have a significant influence on patterns of thought? Behavioral vs. MRI scanner  Ch. 6 (Studies 5 and 6) Neuroanatomical Correlates Are individual differences in overall patterns of self-generated thought associated with differences in grey matter concentration and fractional anisotropy? Morphometric neuroimaging (anatomical MRI and DTI) Ch. 7 Necessary Substrate What regions of the brain form the necessary (if not sufficient) neural substrate allowing for the self-generation of thoughts? Neuropsychological lesion studies Ch. 8 Origin and Ontogeny How and where is thought self-generated in the brain? How do initial patterns of activity recruit and interact with other parts of the brain? Human intracranial electrophysiology; magnetoencephalography     	   5	  Why	  study	  the	  self-­‐generated	  thought	  of	  the	  ‘resting’	  brain?	   There are countless reasons to study self-generated thought and how it is instantiated in the central nervous system, but given the difficulty often encountered by the researcher attempting to explain these reasons to peers, colleagues, and funding agencies – and the skepticism with which such explanations are often met – it is worth summarizing some major viewpoints explicitly. At least four major rationales can be articulated, two of them essentially utilitarian, the others of a more idealistic bent. The first reason is the simple ubiquity of self-generated thought: any cognitive activity occurring so frequently and automatically is probably worth investigating and understanding. The second rationale centers on the idea that self-improvement, as well as the effective treatment of mental health disorders involving dysfunctional patterns of thinking or feeling, requires self-knowledge. That is, optimizing or ameliorating our patterns of thought, feeling, and behavior first requires an understanding of what these patterns are and how the brain gives rise to them. The third notion emphasizes that, despite the obvious role of hard work, long study, and intelligence, essentially every truly novel and creative idea throughout history has arisen somehow spontaneously and in a manner beyond the control and understanding even of the thinker (Campbell, 1960, Simonton, 1999a, Simonton, 1999b). Ordinary self-generated thought, on this view, is the infinitely fertile ground from which occasionally spring all the most revolutionary and important ideas and creations of human civilization. Understanding the tails of this distribution of human creativity necessarily requires understanding its center as well, and the many ways in which our brains engage in ‘little c’ creativity in everyday life (Silvia et al., 2014). Finally, a more idealistic view argues simply that the intrinsic functioning of mind and brain is worthy of scientific interest and investigation in its 	   6	  own right, regardless of whether or not any practical benefits follow. Each of these views is considered in more detail in the following sections.  The	  ubiquity	  of	  self-­‐generated	  thought	  A central impetus for the increased attention now being given to the brain basis of self-generated thought was the early recognition of its sheer quantity. Early researchers reported, and contemporary research continues to confirm, that self-generated thoughts occur ubiquitously in everyday life (Killingsworth and Gilbert, 2010), in laboratory settings (Stawarczyk et al., 2011a), and across cultures (Song and Wang, 2012). Moreover, self-generated thought is prevalent not just during minimally-demanding tasks, or periods of rest or boredom, but during virtually every imaginable activity, from personal grooming to love-making (Killingsworth and Gilbert, 2010). The sheer ubiquity of self-generated thought therefore suggests some importance to normal cognitive function, and recommends it as an object worthy of study. 	  Self-­‐knowledge,	  self-­‐improvement,	  and	  mental	  health	  What we might call the meliorist viewpoint is epitomized by the famous words that were allegedly inscribed in the forecourt to the Temple of Apollo at ancient Delphi: γνωθι σεαυτόν (gnothi seauton, ‘know thyself’) (Pausanius, 1964). A meliorist stance would argue that self-improvement (for the healthy) or recovery (for those with mental health conditions) depends critically upon self-knowledge. Suboptimal or pathological patterns of thinking and behavior need first to be understood before they can be ameliorated and overcome. Presupposed in this view is the notion that cognitive-affective functioning in most, perhaps all, people is suboptimal and contains room for improvement (Fox et al., 2016) – an argument with which it is difficult to 	   7	  disagree. This viewpoint is incredibly pervasive throughout many disparate facets of modern culture, underlying (tacitly or explicitly) many contemplative traditions emphasizing observation of the mind (Ouspensky, 1957, Anālayo, 2003, Kabat-Zinn, 2003, Gunaratana, 2011), as well as most psychiatric, psychoanalytic, and cognitive-behavioral therapeutic schools of thought (Rothbaum et al., 2000, Guntrip, 2008, Kohut, 2013). Indeed, so pervasive has this idea become that it is easy to forget that the high value we tend to place on self-knowledge and basic meta-awareness is essentially a cultural convention, and that self-reflective awareness of the workings of one’s own mind might have been much less prevalent, and prized, in the past – perhaps even altogether absent (Jaynes, 1976, Kuijsten, 2006). Increasing recognition of the relevance of individual patterns of thinking to everyday well-being at the center of the spectrum, and to mental health disorders and spiritual development at its extreme poles, means that there is an increasing need to understand both the phenomenological content and neural substrates of self-generated forms of cognition. Understanding what ‘normal’ people think about, and how their brains generate such thoughts, is a useful (we would argue essential) starting point for understanding how self-generated thought can go awry, for instance in depressive rumination (Berman et al., 2010), involuntary re-experiencing in post-traumatic stress disorder (Ehlers et al., 2004, Berntsen and Rubin, 2008), and dementia (Irish et al., 2012). A basic grasp of the psychology and neurophysiology of self-generated thought has obvious relevance for spiritual paths as well (Brewer et al., 2011, Schooler et al., 2014, Eifring, 2015), but such a discussion is beyond the scope of the present work. The clinical relevance of a basic neuroscientific and neurobiological understanding of self-generated thought, however, is a theme that will recur throughout the ensuing discussions.  	   8	  Creativity,	  insight,	  and	  self-­‐generated	  thought	  A second view credits self-generated and spontaneous forms of thought with virtually every important insight and creation throughout the history of the sciences and the arts. While the role of deliberate study, practice, and thought are of course acknowledged, the argument runs that the key insights and truly novel recombinations of ideas are, somewhat mysteriously, generated ‘in’ us but not ‘by’ us; we are the authors, but perhaps not the agents, behind our own inspirations and creative output (Campbell, 1960, Simonton, 1999a, Simonton, 1999b). Schopenhauer aptly described this process when he insisted that mental ‘rest’ must be cultivated, and space made in which great thoughts can arise of their own accord (Schopenhauer and Payne, 1974). He wrote, “Sound serious thoughts on worthy subjects... cannot be conjured up arbitrarily and at any time. All we can do is to keep the path clear for them... We need only keep the field open to sound ideas and they will come. Therefore whenever we have a free moment with nothing to do, we should… let our mind become tranquil, and then in it something good may easily arise” (pg. 54). To be sure, novel ideas are both facilitated and constrained by our external environments, but nonetheless the remarkable yet totally commonplace creative capacity of the mind-brain needs to be recognized and understood. On this view the creations of artistic and scientific genius are simply the extremely successful examples of the kinds of thought all of us are generating virtually all of the time. We all generate many novel thoughts, images, and plans on a daily basis (Silvia et al., 2014), and an understanding of ‘Creativity’ requires an understanding of the brain’s basic capacity for autonomously generating novel content independent of its environment – despite the fact that much of this self-generated content is rather mundane.  	   9	  Pure	  scientific	  inquiry	  and	  self-­‐generated	  thought	  What could be called the scientific reductionist viewpoint would argue that the mind and brain, like every other part of the universe, is inherently worth investigating and understanding, and that the best initial approach to such an understanding is isolation of the object of interest. ‘The’ scientific method is an ever-evolving and culturally-defined means of inquiry (Bunge, 1998), but nonetheless a series of principles of scientific investigation has long been articulated and embraced by a majority of researchers, whether explicitly or implicitly. The classic reductionist scientific approach has been to isolate the system of interest as much as possible and vary one parameter at a time in order to begin to grasp the functioning and qualities of a given object or organism. Despite the practical and philosophical limitations of this program, there is no doubt that it has been extremely successful in expanding human knowledge in virtually every field of inquiry (Capra and Luisi, 2014). Another rationale for studying the self-generated activity of the (mostly) isolated brain, then, is that observation of the mind in relative isolation might reveal much about its intrinsic functioning and proclivities. The physician John C. Lilly sought to apply this methodology to the most complex object in the known universe – the human central nervous system – when he invented the sensory deprivation tank to study the mind in isolation from its surrounding environment and interactions with others (Lilly, 1956, 1972, Lilly, 1977). Lilly explained: “During those times when it [the mind] is unencumbered by the necessities of interlock with… an external reality, its noninterlock structure can be studied. A given mind seen in pure culture by itself in profound physical isolation and in solitude is the raw material for our investigation” (Lilly, 1972, pg. xxii). This viewpoint, then, would argue that studying the mind-brain in a largely ‘unperturbed’ state is of inherent scientific interest merely for its own sake. 	   10	   The corollary to all this is that the potential benefits and insights of pure research are impossible to anticipate and appreciate. Pure scientific inquiry into self-generated thinking and its brain basis might very well shed light on basic principles of central nervous system functioning or personality psychology, but such an outcome is neither necessary nor promised. An oft-repeated anecdote regarding the discovery of the relationship between electricity and magnetism by Michael Faraday (though perhaps apocryphal) aptly illustrates the point: After demonstrating that electric current created a magnetic field which could exert motive force on susceptible objects within the field (such as a nearby compass needle), a questioner allegedly asked Faraday to what possible use this quaint discovery could ever be put. Faraday is alleged to have responded: “Sir, of what use is a newborn baby?” Analogously, we might contend that the infancy of cognitive neuroscience as a field should caution us against any race to apply our meager understanding of intrinsic patterns of mental and neurophysiological functioning in quests to cure mental illness, explain personality, and so on. That is, we should study self-generated thought and its brain basis because we can, and because it is inherently interesting. 	  The	  content	  of	  self-­‐generated	  thought	   There are many sensory and more abstract qualities of thought that have been investigated by psychologists over the past few decades. We briefly review this content here to give a sense of the vast diversity of thought. Moreover, familiarity with recurring themes in the content of thought will help not only to motivate, but to interpret, neuroscientific investigations of self-generated thought: a detailed and ever-improving understanding of the content of self-generated thought helps mitigate the problem of attributing functions to these areas via reverse inference (Christoff and Owen, 2006, Poldrack, 2006). We now know, from dozens of content 	   11	  studies over the past six decades, that self-generated thought content is largely audiovisual in nature (Klinger and Cox, 1987, Delamillieure et al., 2010, Stawarczyk et al., 2011a, Song and Wang, 2012); is accompanied by a wide variety of emotions (Killingsworth and Gilbert, 2010, Fox et al., 2014, Krans et al., 2015); often involves interoceptive signals and other thoughts about the body (Delamillieure et al., 2010, Diaz et al., 2013); and is typically structured as narratives involving imagined social interactions (Mar et al., 2012) centering on an individual’s current concerns, plans, and goals (Diaz et al., 2013, Klinger, 2013, Poerio et al., 2013, Stawarczyk et al., 2013). These content studies exploring the phenomenological experience of mind-wandering and other kinds of self-generated thought can therefore inform efforts to understand how diverse brain regions might be recruited in association with specific types of content, which we explore in later chapters.  Implausibility and bizarreness are also common features of self-generated thought in waking and also sleep. Although many thoughts involve relatively realistic simulations of plausible events in the external world, nonetheless a substantial number (~20% of reports) contain elements that are bizarre, implausible, or fanciful (defined as ‘departing substantially from physical or social reality’) (Klinger and Cox, 1987, Kroll-Mensing, 1992, Klinger, 2008). A more recent study has provided a general replication of earlier results: analyzing thousands of thoughts reported by 124 subjects, Kane and colleagues found that the average self-generated thought contained a moderate level of fantasy (a mean of 3.77 on a 7-point scale) (Kane et al., 2007). Temporal orientation has also been investigated. Self-generated thought appears to involve roughly equal percentages of thoughts about the past and future (Fransson, 2006), though some studies suggest a ‘prospective bias’ toward future-oriented thoughts (Smallwood et al., 	   12	  2009, Andrews-Hanna et al., 2010, Stawarczyk et al., 2011a), and also a past-bias inducible by negative mood (Smallwood and O'Connor, 2011). Overall, however, it is clear that memories, particularly episodic ones, play a large role in self-generated thought. Many studies have reported a high prevalence (~20% or more of reports) of past-focused mind-wandering (Fransson, 2006, Smallwood et al., 2009, Andrews-Hanna et al., 2010, Smallwood and O'Connor, 2011). Indeed, one of the first studies to explore ‘resting state’ activity using PET noted the similarities between such activity and episodic memory recall, as well as the fact that subjective reports of ‘rest’ actually involved a large amount of past recollection and future planning (Andreasen et al., 1995). Regardless of sensory modality or temporal orientation, the content of self-generated thought centers heavily on people’s current goals and concerns in life (Klinger and Cox, 1987, Klinger, 2008, McMillan et al., 2013). Further, the large percentage of self-generated thoughts centered around the present time ±1 day (~40% in one recent study: Andrews-Hanna et al., 2010) supports the notion that thought centers on current concerns and experiences. Future-oriented thought is also incredibly common during MW (Smallwood et al., 2009, Andrews-Hanna et al., 2010, Stawarczyk et al., 2011a), further supporting a role for self-generated thought in future-planning and potentially problem-solving.  Mentalizing and simulated social interaction are also key components. Self-generated thought often involves imagined social interactions with others, as well as thoughts about the intentions and beliefs of other people (Klinger, 2008, Mar et al., 2012). This has led to the general notion that ‘mentalizing’ (i.e., thinking about the thoughts and minds of others) and social problem-solving via the consideration of hypothetical social situations may be key functional roles of self-generated thought (Buckner et al., 2008b, Andrews-Hanna, 2012, 	   13	  Andrews-Hanna et al., 2014). Supporting this idea, numerous studies have found that brain activity underlying ‘theory of mind’ and mentalizing overlaps significantly with the same regions and networks implicated in self-generated thought (Spreng et al., 2009, Fox et al., 2015).   The	  valence	  of	  self-­‐generated	  thought	   Affect is ubiquitous during self-generated thought. One recent study, for instance, involving thousands of reports, found that the majority (69%) of self-generated thought reports involved emotion (positive emotion in 42.5% of reports, negative emotion in 26.5%), whereas only 31% of reports were reported to be emotionally neutral (Killingsworth and Gilbert, 2010). A prevalent viewpoint has been that self-generated thought is, overall, characterized by unhappiness or negative mood (Killingsworth and Gilbert, 2010, Wilson et al., 2014). Surprised by these claims, we recently reviewed all investigations of self-generated thought that examined affective valence, and found that in fact all studies to date have reported a mild positivity bias to self-generated thought (Fox et al., 2014, Fox et al., in preparation). This general positivity bias (Fig. 1) should not overshadow the fact that variability is high both between and within individuals, and that an incredible diversity of affect can be experienced in self-generated thought (Fox et al., 2014, Fox et al., in preparation). 	   14	  	  Figure	  1.	  Average reported affect across 9 studies investigating the valence of self-generated thought. Note that regardless of the particular scales, anchor points, terms, or questions employed, average valence (red dots) is always mildly positive (slightly above the midpoint on every scale). Reproduced from Fox et al. (in preparation). 	  	   	  	   15	  Cognitive	  control	  and	  metacognition	  in	  self-­‐generated	  thought	  A lack of specific goals, and an unawareness that one is even daydreaming or has deviated from the task at hand, are typical of self-generated thought (Schooler et al., 2011, Fox and Christoff, 2015). But although self-generated thought tends to be less characterized by intentional direction and guidance and self-reflective awareness, this is not always the case. One’s impression of the ‘controllability’ of a segment of self-generated varies widely, from a sense of being able to end it at any time, to being completely absorbed and swept along (Klinger, 1978, Klinger and Cox, 1987, Kroll-Mensing, 1992, Klinger and Kroll-Mensing, 1995, Klinger, 2008). Waking self-generated thought also features a high proportion of goal-related thoughts (~50% of all thoughts; see Chapters 4 to 6), a high degree of future planning (Baird et al., 2011), a moderate (on average) amount of intentional direction of thought flow and content (~25% or more of thoughts; see Chapters 4 to 6), and relatively high levels of meta-awareness  (~44% of thoughts; Christoff et al., 2009). Waking self-generated thought in general therefore involves an appreciable amount of executive and metacognitive functioning, consistent with recruitment of canonical control network regions (see Chapter 2). Collectively, these results suggest that cognitive control and metacognitive awareness in waking self-generated thought lie somewhere between the relative lucidity and self-reflectiveness of externally-directed thought and behavior, and the near-total lack of control and metacognitive nescience characteristic of cognition in sleep and dreaming (see Chapter 3). A major aim of subsequent chapters (especially Chapters 4 and 6) is to elucidate neural correlates and mechanisms that support such executive and metacognitive operations in self-generated thought.  	   16	  CHAPTER	  2	  –	  STUDY	  1:	  GENERAL	  NEURAL	  CORRELATES	  OF	  SELF-­‐GENERATED	  THOUGHT:	  A	  META-­‐ANALYTIC	  ASSESSMENT	  OF	  FUNCTIONAL	  NEUROIMAGING	  INVESTIGATIONS	   Introduction	  In parallel with the burgeoning investigation of self-generated thought, a large and influential literature has accumulated exploring the so-called ‘default mode’ network (DMN). The DMN consists of a number of functionally and anatomically interconnected brain regions that show greater activation at rest than during a variety of externally-oriented attention tasks (Shulman et al., 1997, Buckner et al., 2008a). This pattern of activity is thought to reflect a “default state of brain activity” during which “information broadly arising in the external and internal milieu is gathered and evaluated” (Raichle et al., 2001). Given the ample evidence for the occurrence of self-generated thought during ‘rest’ states as well as minimally-demanding tasks (Singer, 1966), a reasonable supposition emerged early on in these investigations that there must be a close connection between DMN regions and mind-wandering (Andreasen et al., 1995, Raichle et al., 2001). Neuroscientific investigations into the brain basis of self-generated thought have since provided substantial empirical support for this link (e.g., (Mason et al., 2007, Christoff et al., 2009, Andrews-Hanna et al., 2010)). The intuitive appeal of this connection, together with the accumulating empirical evidence in its favor, have led cognitive neuroscientists to focus increasingly on the involvement of the DMN in self-generated forms of thought, while at the same time giving relatively little attention to the involvement of non-DMN brain areas. Although many researchers reject a simple equivalence between self-generated thought and DMN activity, in practice the real breadth of brain recruitment during self-generated thought 	   17	  has often been ignored, and a DMN-centric view often shapes the experimental design, analysis, and interpretation of neuroimaging results. Neuroimaging studies of self-generated thought often restrict analyses to DMN regions-of-interest, effectively excluding consideration of other regions. Similarly, neuroimaging studies of clinical disorders marked by dysfunctions of self-generated thought often focus their investigations on changes in DMN functioning, giving little attention to non-DMN areas (Sheline et al., 2009, Berman et al., 2010, Hamilton et al., 2011), and theoretical accounts have adopted a similar stance (e.g., (Anticevic et al., 2012, Whitfield-Gabrieli and Ford, 2012)). This prevalent experimental and theoretical bias severely limits the neuroscientific investigation of the true breadth of self-generated thought and its related clinical conditions. Although the hypothesized connection (Gusnard et al., 2001, Raichle et al., 2001) between self-generated thought processes and activation of the DMN has now been widely validated in cognitive and clinical neuroscience (Andrews-Hanna et al., 2014, Dixon et al., 2014, Fox and Christoff, 2014, Stawarczyk and D'Argembeau, 2015), this relationship has come to overshadow the possible contribution of non-DMN regions to self-generated thought. Empirical neuroimaging work from our own group and our collaborators (Christoff et al., 2004, Christoff et al., 2009, Andrews-Hanna et al., 2010, Christoff, 2012, Ellamil et al., 2016), as well as many other independent research groups (McGuire et al., 1996, Binder et al., 1999, D'Argembeau et al., 2005, McKiernan et al., 2006, Spiers and Maguire, 2006a, Mason et al., 2007, Wang et al., 2009, Dumontheil et al., 2010, Stawarczyk et al., 2011b), has consistently shown recruitment of non-DMN regions during mind-wandering and related forms of self-generated thought. Yet the robustness of these findings has remained largely unacknowledged, and unexamined by systematic whole-brain meta-analysis. As a consequence, activity in a few selected DMN 	   18	  regions-of-interest (ROIs) continues to be used as the standard, and by implication, sufficient neural indicator of self-generated thought in both healthy (Stawarczyk et al., 2011b, Vanhaudenhuyse et al., 2011) and clinical (Berman et al., 2010, Anticevic et al., 2012, Whitfield-Gabrieli and Ford, 2012) populations. Restricting analyses to DMN regions is also problematic because the subjective variety of self-generated thought modes (which can include memory recall, planning for the future, hypothetical imaginings, mentalizing, and so on) suggest correspondingly complex and variegated neural correlates (Andrews-Hanna et al., 2013, Fox et al., 2013, Dixon et al., 2014, Fox and Christoff, 2014); a simple DMN–self-generated thought mapping is unlikely to fulfill this criterion. Here we present a meta-analysis that specifically aimed to assess the consistency of recruitment of various regions throughout the brain during spontaneous thought processes. Rather than focusing on DMN contributions, we took a whole-brain approach to carry out a quantitative meta-analysis of the 24 functional neuroimaging studies of mind-wandering and related self-generated thought processes published to date (see Table 2). The goal is to give due consideration to the involvement of all relevant brain areas and networks.  Methods	  Search	  strategy	  To identify neuroimaging studies related to mind wandering and self-generated thought, we conducted a comprehensive and systematic search of the literature using MEDLINE (http://www.pubmed.com), Google Scholar (http://scholar.google.com), and PsycINFO (http://www.apa.org/pub/databases/psycinfo/index.aspx) for papers containing the words ‘mind-	   19	  wandering’; ‘mind wandering’; ‘spontaneous thought’; ‘stimulus-independent thought’; ‘task-unrelated thought’; or ‘daydreaming,’ from October 2014 to 1996, the year the first explicit functional neuroimaging study of spontaneous thought was published (McGuire et al., 1996). This list of articles was further refined by searching within results for studies that contained any of the words or phrases ‘magnetic resonance imaging’, ‘MRI,’ ‘positron emission tomography,’ ‘neuroimaging,’ or ‘brain’ within the title or abstract. Of the candidate studies, every abstract was read to confirm whether functional neuroimaging methods to study mind-wandering or related spontaneous thought processes was employed (see details of inclusion and exclusion criteria in the next section). The reference lists of each included study, as well as those of several major reviews (Andrews-Hanna, 2012, Christoff, 2012, Andrews-Hanna et al., 2014), were also searched, to ensure completeness.  	  Study	  inclusion	  and	  exclusion	  criteria	  	  	  	  All studies using functional neuroimaging (i.e., functional magnetic resonance imaging (fMRI) or positron emission tomography (PET) were considered, but studies using purely morphometric neuroimaging methods to study mind-wandering (e.g., cortical thickness in (Bernhardt et al., 2014)) were not included. Using the search strategy detailed above, a total of 24 studies were collected (Table 2). Only reports published in peer-reviewed journals were included (results from abstracts, presented talks, etc., were excluded). To be included, studies had to: (i) report specific peak foci of activation in either Talairach or Montreal Neurological Institute (MNI) space; (ii) report group results (case studies of single subjects were excluded); and (iii) involve healthy, non-clinical, ‘normal’ populations. 	   20	  A fourth criterion was employed to ensure that some form of spontaneous thought was taking place: included studies had to (iv) employ some form of retrospective, online, or questionnaire self-report measure that indicated frequency or depth of spontaneous thought processes (a single exception to this criterion was allowed for an early fMRI study, because it was otherwise specifically designed to detect activations associated with spontaneous thought (Christoff et al., 2004); exclusion of this study did not appreciably alter the meta-analytic results, but its inclusion increased the power of our analysis). This criterion avoids the widespread reverse inference that is particularly present in this field (Christoff and Owen, 2006, Poldrack, 2006), where DMN activity is frequently interpreted as indirect evidence of mind-wandering-like processes (in the absence of any reports to that effect from participants themselves). This criterion also allowed for comparison of spontaneous thought frequency across multiple conditions, and therefore for a common feature across all the contrasts and studies included here (despite difference in comparison/control conditions; see below). All contrasts included in this meta-analysis represent a comparison between some state of higher spontaneous thought frequency with a corresponding period or task involving lower spontaneous thought frequency, or, similarly, regions where higher activity predicted higher rates of mind-wandering. Furthermore, all these spontaneous thought frequency ratings are derived directly from questionnaires, experience sampling probes, and so on, rather than based on assumptions about when spontaneous thought frequency should be higher or lower. The mind-wandering measures employed included (a) post-scan retrospective reports (e.g., (McGuire et al., 1996, D'Argembeau et al., 2005); (b) online, trial-by-trial first-person reports of whether spontaneous thoughts were occurring or not (e.g., (Christoff et al., 2009, Stawarczyk et al., 2011b); and (c) questionnaires purporting to measure ‘trait’ levels of fantasy and spontaneous thought (e.g., (Wang et al., 2009). 	   21	  A final method was (d) inferential, in that mind-wandering scores on a given block or task for a group tested only behaviorally were assumed to hold for a separate group of participants tested in the scanner (e.g., (Mason et al., 2007)). Details for each study are provided in Table 2. As a fifth criterion studies had to (v) report whole-brain results. Studies that reported activation foci in ways other than 3D coordinate space were excluded (e.g. ROI analysis) as they are not compatible with the activation likelihood estimation (ALE) analysis approach. Further, ROI findings tend to be biased towards the DMN, reinforcing the false notion that the DMN is specifically (or exclusively) involved in mind-wandering.  Studies satisfying all five criteria (and therefore included in the meta-analysis) are listed in bold font in Table 2. A total of 10 of the 24 studies were included; failure to satisfy criterion (v) accounted for the majority of exclusions (details in Table 2). Two studies (Hasenkamp et al., 2012, Kucyi et al., 2013) met all five criteria, but involved other potentially confounding factors. The first (Hasenkamp et al., 2012) employed a population of expert meditation practitioners, who may have altered frequency of mind-wandering (Mrazek et al., 2013), as well as altered brain activity during periods of rest with high levels of mind-wandering (Brewer et al., 2011). The second study (Kucyi et al., 2013) examined the relationship between mind-wandering and pain, administrating painful electrical nerve stimulation during the sessions where mind-wandering was reported, again presenting a potential confound. To avoid these confounds affecting our meta-analytic results these studies were excluded (although their inclusion did not appreciably impact the present findings).  One study by Spiers and Maguire (2006a) was followed up with a similar study using the same subjects, dataset, and baseline condition, but somewhat different spontaneous mental activity as the condition of interest (Spiers and Maguire, 2006b). As inclusion of both studies 	   22	  would involve completely overlapping, non-independent samples and data, only one study was included. The former study involved spontaneous mentalizing, which is very common during mind-wandering and related forms of spontaneous cognition (Fox et al., 2013), and so was retained. The latter study instead investigated spontaneous route-planning during virtual navigation, which bears much less resemblance to the various other forms of spontaneous thought examined here, and so was not included in the meta-analysis. 	  Adjustment	  of	  peak	  coordinates	  	  Some included studies reported results in Talairach coordinates. For consistency, we converted all coordinates to MNI space using non-linear transformations in the WFU PickAtlas software package (Maldjian et al., 2003). A small but substantial number of peak foci (n = 16) were reported at coordinates that lay outside of the template brain images used within the activation likelihood estimation meta-analysis software, GingerALE. To avoid data loss, we performed a minimum linear translation of these peak coordinates in MNI space to fit within the GingerALE template mask. In all cases these translations were minimal (mean = 3.69 mm; range: 1–10 mm), and in no case did adjusted peak foci subsequently fall within a different brain region from that listed in the results tables of the original publication. Full details of these translations, including the original and adjusted foci, are available in the online supplementary materials accompanying the original publication (Fox et al., 2015). 	  Activation	  likelihood	  estimation	  (ALE)	  meta-­‐analysis	  We used a quantitative, random-effects meta-analytic method known as activation likelihood estimation (ALE) (Turkeltaub et al., 2002, Laird et al., 2005, Eickhoff et al., 2009, 	   23	  Eickhoff et al., 2012, Turkeltaub et al., 2012) implemented in the software program GingerALE 2.3.1 (San Antonio, TX: UT Health Science Center Research Imaging Institute). The most recent ALE algorithm tests for above-chance clustering of peak foci from different experiments included in the meta-analysis (Eickhoff et al., 2009, Eickhoff et al., 2012) by comparing actual activation foci locations/clustering with a null distribution that includes the same number of peak foci distributed randomly throughout the brain’s grey matter. Included activation foci were smoothed using a full-width half maximum (FWHM) Gaussian kernel dependent on the sample size (subjects) of the experiment from which foci were drawn (larger sample -> smaller smoothing kernel – empirically determined by (Eickhoff et al., 2009, Eickhoff et al., 2012)). Resulting statistical maps show clusters where convergence between activation foci is greater than would be expected by chance (i.e., if foci from each experiment were distributed independently). We analyzed a total of 157 foci from 10 neuroimaging studies (studies in bold font in Table 2). Statistical maps were thresholded using a false discovery rate (FDR – see (Genovese et al., 2002)) of q = .05 and a cluster threshold of k = 1000 mm3. We used this relatively liberal statistical threshold, but high cluster threshold, in order to better visualize results and to ensure that cross-study convergence was not neglected in this relatively small-sample meta-analysis. Similar results were found at more conservative statistical thresholds. Region classifications follow those indicated in the Multi-Image Analysis GUI (‘Mango’) image-viewing software (UT Health Science Center Research Imaging Institute) used to visualize the meta-analytic results. The Duvernoy neuroanatomical atlas was also consulted (Duvernoy et al., 1991). 	   	  	   24	  Table	  2.	  Functional neuroimaging studies examining self-generated thought.	  Study Analysis Peak foci Design N Mind-wandering measurea Thought type McGuire et al., 1996 WB & ROI 5 Blocked 5/6 Retrospective Verbal stimulus-independent thoughts during rest Binder et al., 1999 WB 8 Blocked 14 Inferential Task-unrelated thought during rest Christoff et al., 2004 WB 15 Blocked 12 Assumptive Spontaneous thought during rest D’Argembeau et al., 2005 WB 9 Blocked 13 Retrospective Spontaneous thought during rest McKiernan et al., 2006 ROI – Blocked 30 Inferential  Task-unrelated thought during auditory task Spiers & Maguire, 2006 WB & ROI 24 ER 20 Retrospective (Online) Spontaneous mentalizing during navigation task Mason et al., 2007 WB resting state à Functionally-defined ROIs 20 Blocked 19 Inferential and Questionnaire Mind-wandering during highly practiced task Christoff et al., 2009 WB 34 ER 15 Online Mind-wandering during SART Wang et al., 2009 WB & ROI 8 Blocked 13 Questionnaire Spontaneous thought during rest Andrews-Hanna et al., 2010 ROI; seed-based FC – Blocked  30/ 139 Surprise retrospective Spontaneous thought during passive fixation Dumontheil et al., 2010 WB 20 Blocked 16 Retrospective Task-unrelated thought during various simple tasks Stawarczyk et al., 2011 ROI/Supp. WB 14 ER 22 Online Mind-wandering during SART Vanhaudenhuyse et al., 2011 ROI – ER 22 Online Intensity of internal awareness during rest Christoff, 2012 ROI; seed-based FC – ER 15 Online Mind-wandering during SART Hasenkamp et al., 2012 WB – ER 14 Online Mind-wandering during meditation Hasenkamp & Barsalou, 2012 ROI; seed-based FC – ER 14 Online Mind-wandering during meditation Allen et al., 2013 ROI – ER 21/ 21 Online Task-unrelated thought during EAT Kucyi et al., 2013 WB – Blocked 51 Online Mind-wandering during painful stimuli Moss et al., 2013 Functional localizer-defined ROIs – Blocked 15 Retrospective Mind-wandering during reading Smallwood, Tipper, et al., 2013 WB – ER 16 Assumptive Stimulus-independent thought during various tasks Smallwood, Gorgolewski, et al., 2013  ROI – Blocked 42 Online Mind-wandering during reading Gorgolewski et al., 2014 WB – Blocked 166 Retrospective Self-generated thought during rest Kucyi & Davis, 2014 ROI – Blocked 51 Online Mind-wandering during painful stimuli Tusche et al., 2014 ROI – ER 30 Online Self-generated thought during rest All studies were conducted using fMRI, except McGuire et al. (1996) and D’Argembeau et al. (2005), which used PET. N = sample size (experimental/control group, if applicable). Studies employing whole-brain analyses that were 	   25	  included in the meta-analysis are listed in bold font. The number of peak activation foci contributed from each study is indicated in the third column. EAT: Error Awareness Task; ER: event-related; FC: functional connectivity; ROI: region of interest; SART: Sustained Attention to Response Task; Supp.: supplemental; WB: whole-brain. a See Methods for a discussion of mind-wandering measures.  	  Results	  	   We found 13 regions to be reliably involved in mind wandering and self-generated thought (Table 3; Fig. 2). These regions included all major hubs of the DMN: rostromedial prefrontal cortex, medial prefrontal cortex/anterior cingulate cortex, precuneus/posterior cingulate cortex, bilateral inferior parietal lobule, and left medial temporal lobe/parahippocampal cortex (extending somewhat into the cerebellum). We also found significant clusters outside the DMN, including dorsal anterior cingulate cortex, right rostrolateral prefrontal cortex, left ventrolateral prefrontal cortex, secondary somatosensory cortex, left temporopolar cortex, left mid insula, and the left lingual gyrus.        	   	  	   26	  Table	  3.	  Meta-analytic results: brain areas consistently activated by mind-wandering and related self-generated thought processes. Region	   Cluster	  Size	  (mm3)	  Peak	  ALE	  Value	  Peak	  in	  MNI	  Space	  (x,	  y,	  z)	  	  Brodmann	  Area	  Studies	  Contributing	  to	  Cluster	  Frontal	  lobe	  Dorsal	  anterior	  cingulate	  cortex	   2920	   0.0144	   -­‐5,	  27,	  39	  (Fig.	  2a)	   32	   a,	  e,	  g,	  h	  R	  lateral	  prefrontal	  cortex	   1656	   0.0085	   45,	  43,	  -­‐8	  (Fig.	  2e,g)	   10/46	   d,	  i	  Rostromedial	  prefrontal	  cortex	   1256	   0.0081	   3,	  61,	  13	  (Fig.	  2b)	   10/9	   e,	  f	  Medial	  prefrontal	  cortex;	  anterior	  cingulate	  cortex	   1080	   0.0089	   4,	  42,	  3	  (Fig.	  2b)	   24/32	   f,	  g	  L	  ventrolateral	  prefrontal	  cortex	   1032	   0.0126	   -­‐42,	  29,	  -­‐12	  (Fig.	  2d)	   47/11	   e,	  j	  Parietal	  Lobe	  Precuneus;	  posterior	  cingulate	  cortex	   3784	   0.0123	   -­‐8,	  -­‐56,	  39	  (Fig.	  2a)	   7/31	   b,	  d,	  f,	  g,	  h,	  j	  L	  inferior	  parietal	  lobule;	  angular	  gyrus	   2616	   0.0092	   -­‐46,	  -­‐72,	  25	  (Fig.	  2d)	   39	   b,	  e,	  f,	  g,	  h	  R	  inferior	  parietal	  lobule;	  supramarginal	  gyrus	   1808	   0.0098	   56,	  -­‐51,	  33	  (Fig.	  2f)	   40/39	   d,	  f,	  i	  R	  secondary	  somatosensory	  cortex	   1416	   0.0097	   24,	  -­‐39,	  56	  (Fig.	  2i)	   5/40	   c,	  e,	  g	  Temporal	  Lobe	  L	  parahippocampus	   3496	   0.0196	   -­‐27,	  -­‐37,	  -­‐18	  (Fig.	  2h)	   36	   b,	  c,	  f,	  g,	  h	  L	  mid	  insula/superior	  temporal	  gyrus	   2608	   0.0116	   -­‐50,	  -­‐1,	  -­‐5	  (Fig.	  2d,g)	   13/22	   f,	  g,	  i	  L	  temporopolar	  cortex	   1320	   0.0090	   -­‐35,	  10,	  -­‐26	  (Fig.	  3)	   38	   c,	  g,	  j	  Occipital	  Lobe	  L	  lingual	  gyrus	   1936	   0.0092	   -­‐15,	  -­‐66,	  5	  (Fig.	  2c)	   19/18/30	   c,	  e,	  g,	  j	  Significant meta-analytic clusters, with peak coordinates, for self-generated thought > control conditions. BA: Brodmann area. a: McGuire et al., 1996; b: Binder et al., 1999; c: Christoff et al., 2004; d: D’Argembeau et al., 2005; e: Spiers & Maguire, 2006; f: Mason et al., 2007; g: Christoff et al., 2009; h: Wang et al., 2009; i: Dumontheil et al., 2010; j: Stawarczyk et al., 2011.   	   27	  	  Figure	  2.	  Peaks of each meta-analytic cluster of activation associated with self-generated thought processes. Peaks for each of 13 significant meta-analytic clusters of activation associated with mind-wandering and related self-generated thought processes. (a) Posterior cingulate cortex/precuneus and dorsal anterior cingulate cortex. (b) Medial prefrontal cortex/anterior cingulate cortex and rostromedial prefrontal cortex. (c) Left lingual gyrus (largest cluster). (d) From left to right: left inferior parietal lobule/angular gyrus; left mid-insula; and left ventrolateral prefrontal cortex. (e) Right rostrolateral prefrontal cortex. (f) Right inferior parietal lobule/supramarginal gyrus. (g) Left temporopolar cortex and mid-insula; right rostrolateral/dorsolateral prefrontal cortex. (h) Left parahippocampus/medial temporal lobe, extending into the cerebellum. (i) Right secondary somatosensory cortex. Color bar indicates ALE likelihood values (see Methods). Stereotactic coordinates (x, y, z) are in MNI space. L: left; R: right.	     	   28	  	  Figure	  3	  Meta-analytic activations for self-generated thought. Cortical mapping of significant meta-analytic clusters associated with mind-wandering and related self-generated thought processes (green clusters) juxtaposed with outlines of the default mode network (blue) and the frontoparietal control network (red). Note that spontaneous thought activations overlap considerably with both networks, but also include regions beyond both networks (highlighted in Figure 1). Default mode network and frontoparietal control network masks based on Yeo et al. (2011). Reproduced from Fox et al. (2015).   Discussion	  What	  constitutes	  the	  default	  mode	  network?	  Understanding the extent to which self-generated thought recruitment overlaps with, and extends beyond, the DMN naturally requires a definition of what brain areas constitute the DMN. Because the DMN is functionally (rather than anatomically) defined, and because numerous 	   29	  different methods can be used to derive this functional definition, its boundaries vary somewhat depending on the methods and contrasts employed. Early studies using positron emission tomography compared blocked periods of ‘rest’ with blocked task periods (Andreasen et al., 1995, Shulman et al., 1997, Raichle et al., 2001). Based on these studies, the DMN came to be defined as a set of regions exhibiting decreases in activity, or deactivations, during a variety of externally-directed tasks compared to periods of rest (Shulman et al., 1997, Mazoyer et al., 2001, Raichle et al., 2001, Buckner et al., 2008b). Later work using briefer temporal epochs (event-related analyses) suggested a similar constellation of regions (Kirchoff et al., 2005), while more recent work has begun to define the DMN based on its pattern of intrinsic resting state functional connectivity (i.e., temporally correlated fluctuations in activity across different brain areas) (Greicius et al., 2003, Yeo et al., 2011). Comparing findings across these three methods has allowed for a tentative delineation of ‘canonical’ DMN regions, which include the medial prefrontal cortex, posterior cingulate cortex, temporoparietal junction (and nearby inferior parietal lobule), and parts of lateral temporal cortex (Buckner et al., 2008b, Andrews-Hanna et al., 2014). Other regions are also sometimes included, most commonly the medial temporal lobe (especially the hippocampus) and ventrolateral prefrontal cortex (BA 47) (Buckner et al., 2008b, Andrews-Hanna et al., 2014).  The	  default	  mode	  network	  and	  self-­‐generated	  thought	  Since the beginning of investigation into DMN functions, this network has been linked to mental processes such as spontaneous self-referential processing, memory recall, and inner simulations of behavior (Andreasen et al., 1995, Gusnard et al., 2001, Raichle et al., 2001). A 	   30	  growing body of work focusing specifically on self-generated thought has bolstered this connection: both retrospective (Mason et al., 2007, Andrews-Hanna et al., 2010) and online, trial-by-trial (Christoff et al., 2009) self-reported mind-wandering predict increased activity in DMN regions. Further, self-reported depth (or ‘intensity’) of internally-directed thought predicts higher activation in DMN areas (Vanhaudenhuyse et al., 2011); and spontaneous mentalizing (i.e., mind-wandering about the thoughts or intentions of others) also recruits the DMN (Spiers and Maguire, 2006a).  Regardless of the specific DMN borders used, a strong connection between the DMN and spontaneous thought emerges. Indeed, all studies of self-generated thought published so far employing an agnostic, whole-brain approach (McGuire et al., 1996, Binder et al., 1999, Christoff et al., 2004, D'Argembeau et al., 2005, Spiers and Maguire, 2006a, Mason et al., 2007, Christoff et al., 2009, Wang et al., 2009, Dumontheil et al., 2010, Stawarczyk et al., 2011b) have shown recruitment of at least some DMN regions, including medial prefrontal cortex (BA 10/32), posterior cingulate cortex/precuneus (BA 7/31), parahippocampal cortex (BA 36), and temporoparietal junction/inferior parietal lobule (BA 39/40). Our quantitative meta-analysis corroborates these findings, strongly supporting a link between the DMN and self-generated thought (Table 3; Fig. 2 and 3). Putative roles for DMN hubs and subsystems in self-generated thought are discussed at length elsewhere (Andrews-Hanna et al., 2014, Smallwood and Schooler, 2014). 	  Frontoparietal	  control	  network	  involvement	  in	  self-­‐generated	  thought	  Mind-wandering and related self-generated thought processes consistently recruited many regions of the frontoparietal control network (Vincent et al., 2008, Yeo et al., 2011), including the dorsal anterior cingulate cortex (BA 32), right rostrolateral prefrontal cortex (BA 10/46), 	   31	  right anterior inferior parietal lobule (BA 39/40), and precuneus (BA 7) (Table 3; Figs. 2 and 3). The frontoparietal control network plays a central role in goal-directed cognition (Niendam et al., 2012). This role extends from classic measures of executive functioning to personally relevant goal-directed tasks, such as planning for the future (Spreng et al., 2010, Gerlach et al., 2014). In the latter capacity, it tends to be activated in unison with DMN areas, so we discuss its potential role in spontaneous thought in the following section, which addresses DMN-frontoparietal control network coupling in the generation of thought.  Coupling	  of	  the	  default	  mode	  and	  frontoparietal	  control	  networks	  Many meta-analytic clusters fell within the borders of either the default mode (Shulman et al., 1997, Raichle et al., 2001, Buckner et al., 2008b) or fronto-parietal control network (Vincent et al., 2008, Yeo et al., 2011, Spreng et al., 2013). We visualize the extent of this overlap by plotting our meta-analytic clusters on an ‘inflated’ brain template with border outlines for both networks in Figure 3. The borders of the two networks are based on previously published, large-scale (n = 1000) resting state datasets (Yeo et al., 2011), whereas the green clusters of activation represent our whole-brain meta-analytic results (from Table 3; Fig. 2).  Although our meta-analytic results cannot directly address the temporal correlation of these brain activations, such results are consistent with observations of temporal coupling of DMN and fronto-parietal control network areas (e.g., (Spreng et al., 2010, Christoff, 2012)). Self-generated thought is not the only cognitive process, however, to exhibit co-activation of, and functional connectivity (i.e., temporal coupling) between, the default mode and frontoparietal control networks. Similar findings have been observed in autobiographical memory recall (Spreng et al., 2009), mentalizing or ‘theory of mind’ (Spiers and Maguire, 	   32	  2006a, Spreng et al., 2009), creative generation and evaluation of artwork (Ellamil et al., 2012), and goal-directed prospection (Spreng et al., 2009, Gerlach et al., 2014). Intriguingly, memory recall, mentalizing, creative recombination of ideas, and thoughts about the future mark the dominant themes of self-generated thought content from the subjective point of view (reviewed in (Klinger, 2008, Fox et al., 2013)), making these overlapping brain activations intriguing avenues for future research. Resting-state functional connectivity analysis has revealed that there is an extensive intrinsic architecture to support dynamic interactions between the default and frontoparietal control networks (Spreng et al., 2013). The coupled recruitment of these two networks might represent a dynamic interplay whereby executive control regions guide, evaluate, and select among the various self-generated streams of thoughts, memories, and imaginings offered up to consciousness by the DMN (Andrews-Hanna et al., 2014, Fox and Christoff, 2014, Fox et al., 2016).  Other	  cortical	  areas	  recruited	  by	  self-­‐generated	  thought	  Areas outside the default and frontoparietal control networks also need to be given consideration. The temporopolar cortex (McGuire et al., 1996, Christoff et al., 2004, Christoff et al., 2009, Dumontheil et al., 2010), the insula (Christoff et al., 2004, Mason et al., 2007, Christoff et al., 2009, Dumontheil et al., 2010, Stawarczyk et al., 2011b), and the lingual gyrus/medial occipital cortex are all relatively consistently activated during self-generated thought (Fig. 2), but do not clearly belong to either of these two networks. Further, our meta-analysis also suggests consistent recruitment of secondary somatosensory cortices (Fox et al., 2015). What role could these regions be playing? 	   33	  A detailed and ever-improving understanding of the content of self-generated thought helps mitigate the problem of attributing functions to these areas via inverse inference (Christoff and Owen, 2006, Poldrack, 2006), allowing some informed (if preliminary) hypotheses to be put forward. We now know, from dozens of content studies over the past six decades, that spontaneous thought content is largely audiovisual in nature (Klinger and Cox, 1987, Delamillieure et al., 2010, Stawarczyk et al., 2011a, Song and Wang, 2012); is accompanied by a wide variety of emotions (Killingsworth and Gilbert, 2010, Fox et al., 2014, Krans et al., 2015); often involves interoceptive signals and other thoughts about the body (Delamillieure et al., 2010, Diaz et al., 2013); and is typically structured as narratives involving imagined social interactions (Mar et al., 2012) centering on an individual’s current concerns, plans, and goals (Diaz et al., 2013, Klinger, 2013, Poerio et al., 2013, Stawarczyk et al., 2013). These content studies exploring the phenomenological experience of mind-wandering and other kinds of spontaneous thought can therefore inform efforts to understand how diverse brain regions might be recruited in association with specific types of content. For instance, the prominent recruitment of the medial occipital cortical regions, key to visual imagery and the visual component of dreaming (Solms, 1997), are a plausible substrate for the ubiquitous visual imagery in spontaneous thought. Similarly, consistent recruitment of the insula and secondary somatosensory cortex, areas key to interoception (Craig, 2009) and exteroception (Penfield and Boldrey, 1937), respectively, is in line with the prominence of body-related thought content (Delamillieure et al., 2010, Diaz et al., 2013). Temporopolar cortex recruitment could be linked to the ubiquity of conscious emotional experience and mentalizing (Olson et al., 2007), and the prevalence of positive affect in particular (Fox et al., 2014) might be expected to recruit subcortical basal ganglia structures involved in reward (see next section). Finally, the frequency 	   34	  of future-oriented thought (Andrews-Hanna et al., 2010, Stawarczyk et al., 2013) involving planning and goal-related ideation would suggest the recruitment of prefrontal executive areas critical for these capacities – and this is observed in rostrolateral prefrontal cortex, dorsal anterior cingulate cortex, and possibly other lateral prefrontal areas.  These functional roles are largely speculative, of course, and we intend them merely as suggestions. Nonetheless, note that the wide breadth of brain recruitment during spontaneous thought aligns well with the wide variety of phenomenological content. The content dimensions of spontaneous thought are reviewed in more detail by Fox et al. (2013) and Klinger (2008), and further discussion of the possible functional roles of these cortical regions and their relation to content is provided by Domhoff & Fox (2015) and Fox et al. (2015).  Subcortical	  contributions	  to	  self-­‐generated	  thought?	  A corticocentric bias pervades much of cognitive neuroscience (Parvizi, 2009), and this is true too of DMN-centric views of self-generated thought. Empirical reports have shown self-generated thought to recruit numerous subcortical regions, including the cerebellum (Spiers and Maguire, 2006a, Christoff et al., 2009, Stawarczyk et al., 2011b), the putamen and claustrum (Mason et al., 2007), the caudate nucleus (Christoff et al., 2009, Stawarczyk et al., 2011b), the thalamus (Christoff et al., 2009, Stawarczyk et al., 2011b), and the hypothalamus (Stawarczyk et al., 2011b). Although no subcortical clusters emerged in our quantitative meta-analysis (Fox et al., 2015), this could have been for a number of reasons – such as low statistical power due to the small number of studies conducted to date, or because of diversity in the type of thought content investigated in each study. Continuing research may show that certain subcortical structures are consistently recruited during self-generated thought, or are recruited specifically depending on 	   35	  thought content or affect. The recruitment of the basal ganglia, for instance, could be related to the highly automatized and habitual nature of many streams of spontaneous thought (Fox et al., 2016), or possibly the rewarding aspect of the many emotionally positive thoughts reported (Fox et al., 2014). The potential involvement of the cerebellum also raises intriguing possibilities: although traditionally considered mainly a motor area, more recent conceptions see it as playing a critical role in ‘higher’ cognitive processes (Middleton and Strick, 1994, Bellebaum and Daum, 2007, Stoodley and Schmahmann, 2009), especially planning and simulating future behaviors (Houk and Wise, 1995, Thach, 1996, Schmahmann and Sherman, 1998). Cerebellum recruitment might therefore occur preferentially when self-generated thought involves complex planning and simulations. These suggestions are of course speculative at this early stage; the role (if any) played by subcortical brain regions is further explored in our fMRI (Chapter 4) and morphometric MRI (Chapter 6) investigations, below.  Implications	  for	  clinical	  disorders	  of	  self-­‐generated	  thought	  Above, we discussed the idea that the DMN (especially medial temporal lobe structures) may contribute to the actual arising of memories and the stream of associated thoughts and imaginings. Higher executive areas may, on the other hand, subsequently evaluate, monitor, and guide the flow of these thoughts toward desirable and beneficial outcomes (Fox and Christoff, 2014, Fox et al., 2016). Given the evidence that self-generated thought involves a coupling of both DMN and frontoparietal control network regions in normal, healthy subjects, important implications arise for clinical disorders of self-generated thought.  For instance, preferential (or over-) activation of DMN regions in isolation, with minimal or attenuated contributions from executive areas, might characterize dysfunctional 	   36	  forms of self-generated thought, such as depressive rumination or the spontaneous and debilitating re-experiencing characteristic of post-traumatic stress disorder (Fox et al., under review). Conversely, an over-involvement of frontoparietal control areas might characterize conditions like obsessive-compulsive disorder, where fleeting thoughts and urges are acted upon repeatedly and with determination. If the distributed neural basis of spontaneous thought is not appreciated, and DMN regions are preferentially or exclusively investigated, links between such clinical disorders of spontaneous thought and non-DMN brain regions might be missed.  Two highly-cited studies of depression (Berman et al., 2010, Hamilton et al., 2011) exemplify this potential pitfall: both found links between DMN activity and frequency of depressive rumination, but neither explored potential links with non-DMN areas, particularly executive areas of the frontoparietal control network. Yet one of these groups’ meta-analysis (Hamilton et al., 2012) has shown that several regions beyond the DMN show abnormal activity in depression, including the dorsolateral prefrontal cortex, insula, and dorsal anterior cingulate cortex (note that the two latter regions are consistently recruited by self-generated thought; see Results, above). In contrast, a recent study that did explore whole-brain relationships (Hach et al., 2014) found that depressed patients showed altered patterns of functional connectivity across numerous non-DMN areas, including dorsal anterior cingulate cortex, as well as both dorsolateral and rostrolateral prefrontal cortex (all regions of the frontoparietal control network). Although the study investigated deliberate past- and future-oriented thought, and did not directly address ‘spontaneous’ forms of cognition, similar differences in neural recruitment could also characterize the spontaneous thoughts of depressed patients. Such possibilities are clearly worthy of further investigation with an unbiased, whole-brain approach, not only for depression but also other clinical conditions characterized by disorders of spontaneous thought. 	   37	  Limitations	  of	  the	  meta-­‐analysis	   A number of limitations of the present meta-analytic methods, as well as of the primary research literature meta-analyzed, should be kept in mind. First, differences in baselines and other control conditions across studies are a source of heterogeneity. Collapsing across numerous different baselines or control conditions is a common, indeed inevitable, practice in any large meta-analysis of functional neuroimaging studies (Svoboda et al., 2006, Wager et al., 2007, Salimi-Khorshidi et al., 2009, Wager et al., 2009). Nonetheless, the heterogeneity attributable to the many different study designs concatenated here should be kept in mind. Another concern is the wide variety of questionnaire and thought-sampling methods employed to determine the content and frequency of spontaneous thought. Although we would argue that subtle distinctions between different kinds of spontaneous thought are premature at this point, it should be acknowledged that various distinctive forms of spontaneous cognition might eventually be differentiable both phenomenologically, and potentially also at the neural level. The present meta-analysis, collapsing over these potentially distinguishable forms of spontaneous cognition, is meant merely as a preliminary effort at synthesis. A third point is the fairly small sample size of studies included in the meta-analysis. Compared with processes like memory or attention, very little neuroimaging research has investigated spontaneous thought processes, and of this already small pool many studies have not employed agnostic, whole-brain analyses suitable for inclusion in a quantitative meta-analysis. As such, our meta-analytic results are based on a relatively small (if rigorous, and rigorously selected) set of studies, and should therefore be interpreted with caution. Future work may both extend and further clarify the results presented here.  	   38	  Conclusions	  and	  future	  directions	   The primary aim of this meta-analysis was to provide a quantitative overview of the brain regions and networks consistently recruited by mind-wandering and related forms of spontaneous thought. We found that a wide variety of regions throughout all lobes and both hemispheres appear to be reliably recruited. Most of these regions fall within boundaries of the default mode or frontoparietal control networks, suggesting that these networks, as presently conceptualized, form an important part of the neural signature accompanying, self-generated thought of various kinds. A number of other reliably activated regions are not so easily subsumed under either of these networks (e.g., the insula, somatosensory cortices, and the lingual gyrus), suggesting that future work may need to look beyond the typical delineation of these two networks’ contributions in order to construct a full account of the neural recruitment associated with spontaneous thought. The study of deliberate, goal-oriented, and externally-directed mental activity has been of paramount concern to cognitive neuroscientists over the past few decades (Dixon et al., 2014). Although this work has been, and remains, critical in advancing our understanding of brain function, its dominance has led to the relative neglect among researchers of self-generated forms of cognition. To continue without redressing this imbalance in our scientific focus would be to ignore some of the most remarkable and complex processes of which our brains are capable.  A deeper investigation of more fleeting, spontaneous forms of cognition will be necessary to move cognitive neuroscience toward a neurobiological understanding of higher mental functions like creativity and imagination. To paraphrase William James (James, 1985): No account of cognition in its totality can be final which leaves these other forms of consciousness quite disregarded.  	  	   39	  CHAPTER	  3	  –	  STUDY	  2:	  NEURAL	  CORRELATES	  OF	  SELF-­‐GENERATED	  THOUGHT	  IN	  SLEEP	  AND	  DREAMING:	  A	  PRELIMINARY	  META-­‐ANALYTIC	  ASSESSMENT	  OF	  FUNCTIONAL	  NEUROIMAGING	  INVESTIGATIONS	  	  Introduction:	  the	  multiplicity	  of	  sleep	   Awareness of the subjective multiplicity of sleep goes back thousands of years – at least as far as the ancient Indian philosophical texts known as the Upanishads composed around the 6th century B.C.E. (Hume, 1921, Prabhavananda et al., 1984, Deutsch and Dalvi, 2004, Sharma, 2012). Ancient Indian philosophers clearly recognized a distinction between dreamless sleep, dreaming, and even ‘lucid’ dreaming – being aware that one is dreaming while dreaming (Prabhavananda et al., 1984, Sharma, 2012). In the West, Aristotle made strikingly prescient observations for his time: he recognized both dreamless and dreaming sleep; described what we today call sleep-onset hypnagogic imagery; correctly hypothesized that dreaming represents the activity of our perceptual faculties in the absence of external inputs; and even recognized the possibility of lucid dreaming (Aristoteles and Gallop, 1996, Barbera, 2008). And at least one thousand years ago, Tibetan Buddhist practitioners had developed sophisticated cognitive practices oriented toward increasing metacognitive awareness during dreamless sleep and dreaming (Gillespie, 1988, Wangyal, 1998). All of these traditions began a fruitful mapping of quantitative and qualitative psychological (ψ) differences in various mental states throughout the sleep cycle. Western science began finally to put these observations on a firmer footing in the mid-20th century, with the discovery that surface-recorded brain electrical potentials could dissociate between several sleep stages (Aserinsky and Kleitman, 1953, 1955, Dement and Kleitman, 	   40	  1957a, b, Monroe et al., 1965). This research, the first to definitively identify neurobiological (Φ) markers related to particular cognitive states (ψ) during sleep, led to the well-known classification of sleep into rapid-eye-movement (REM) and non-rapid-eye-movement (NREM) stages (Rechtschaffen and Kales, 1968), with four major stages generally recognized by contemporary researchers (NREM1, 2, 3/4, and REM).  These stages have not been equally recognized or researched over the past few decades. REM and NREM (which had not yet been differentiated into substages, and was collectively known as ‘slow wave sleep’ – SWS) were intensively investigated from the beginning of modern sleep and dream science in the 1950’s. It was rapidly recognized that these stages were characterized by differences in subjective experience – most notably, by the high frequency of dream reports following awakening from REM, but the relative paucity of such reports following awakenings from NREM (reviewed by Nielsen, 1999, 2000). A pair of more marginal and difficult to investigate stages were largely ignored until recently: NREM1 (sleep onset) and so-called ‘lucid dreaming,’ in which one is aware of the fact that one is dreaming, while dreaming (LaBerge et al., 1981). Detailed investigation of the electrophysiological substages and phenomenological content of sleep onset (NREM1), although begun in the 1960’s (Foulkes and Vogel, 1965, Foulkes et al., 1966, Vogel et al., 1966), was not conducted in earnest until the 1990’s (Hori et al., 1994, Tanaka et al., 1996, 1997, Hayashi et al., 1999). Lucid REM sleep dreaming (LREM) still remains controversial to many researchers; pioneering but tenuous polysomnographic research from the 1980’s (LaBerge, 1980, LaBerge et al., 1981, Fenwick et al., 1984, LaBerge et al., 1986) has continued to be replicated and extended, however (LaBerge and Levitan, 1995, Holzinger et al., 2006, Voss et al., 2009), as 	   41	  well as investigated with more sophisticated methods, such as combined EEG-fMRI (Dresler et al., 2011, Dresler et al., 2012) and transcranial direct current stimulation (Stumbrys et al., 2013).  Sleep	  mentation	  as	  self-­‐generated	  thought	  We have argued at length elsewhere that mentation during sleep, particularly dreaming per se, can be viewed as an intensified form of waking self-generated thought (Fox et al., 2013, Domhoff and Fox, 2015). The basis for this claim is twofold: both the subjective experience of dreaming, and its neurophysiological correlates (as indexed by REM sleep), strongly parallel those of waking mind-wandering and related forms of self-generated thought.  Waking self-generated thought is typically characterized by auditory and visual imagery, ubiquitous affect, a strong focus on current concerns and social interactions, and varying degrees of narrative structure (Fox et al., 2013, Andrews-Hanna et al., 2014, Fox et al., 2015); see also Chapter 1). The same statements can be made about REM sleep mentation, with the qualification that these characteristics in fact tend to be heightened or exaggerated in dreaming: the audiovisual world is fully present and immersive, emotions more intense and perhaps more ubiquitous, social characters more numerous and interactions with them more elaborate, and narrative structure is extended over time and in more complex ways (Windt, 2010, Fox et al., 2013, Domhoff and Fox, 2015).  A similar parallel is observed at the neurophysiological level. Waking self-generated thought, as compared to active focus on a task or external stimulus, is associated with a relatively consistent pattern of brain activations centered on the default network and extending into medial occipital areas involved in visual imagery, as well as some executive brain regions (see Chapter 2, above). When we previously meta-analyzed functional neuroimaging studies of REM sleep, 	   42	  during which dreaming occurs approximately 80% of the time (Hobson et al., 2000), we found that many of the same brain areas implicated in waking self-generated thought were even more strongly recruited during REM sleep, including medial prefrontal cortex, numerous medial temporal lobe structures, and medial occipital areas (Fox et al., 2013). Additionally, by slightly relaxing statistical thresholds, further overlap in the inferior parietal lobule, another key default network region, was revealed (Domhoff and Fox, 2015).   Overall, these results convinced us that dreaming, and its most common neurophysiological correlate, REM sleep, show an overall intensification or amplification of both the subjective qualities and neural recruitment associated with waking self-generated cognition (cf. Fig. 3 in Fox et al., 2013). Because the NREM sleep stages are also characterized by variable levels of cognitive activity and dream experience, determining their general neural correlates is an attractive target for research, because a general understanding of these neural substrates would allow further examination of the hypothesis that self-generated thought has a common brain basis, regardless of the particular conscious state in which it takes place. Determining the general neural correlates of the NREM sleep stages was therefore the main aim of this preliminary meta-analysis. 	  Sleep	  can	  be	  meaningfully	  dissociated	  into	  stages	  	  The preceding overview only hints at the enormous body of work that has been conducted over the past 60 years within a paradigm whose core assumption is that sleep stages can be meaningfully dissociated and more-or-less independently investigated. Loomis and colleagues (Loomis et al., 1935, Loomis et al., 1937) were the first to provide detailed descriptions of distinct neurophysiological stages in normal human sleep, and much subsequent 	   43	  work has followed, corroborated, and expanded on these efforts. What are the criteria upon which these sleep stages are distinguished – and are they valid?  Nielsen (2014) rightly points out that the widespread use of standard sleep stage scoring criteria (Rechtschaffen and Kales, 1968) has led to an artificially categorical view of sleep stages, accompanied by tacit assumptions of both mutual exclusivity and abrupt transitions. Even the narrow use of just a few electrophysiological markers cannot support such a view; and of the hundreds of potential physiological and neural markers that fluctuate throughout the sleep-wake cycle, only a select few are routinely employed to differentiate sleep stages (Nielsen, 2014). These facts should put us on guard against any facile reification of distinct sleep stages.  A key question therefore needs to be asked: Are sleep stages a fact of neurophysiology or an investigative convenience? The answer, we believe, is that they are somewhere in between. While keeping the above caveats firmly in mind and refraining from reifying classification schemes as actual entities, we agree with the conclusions of various comprehensive reviews of this issue (e.g., (Hobson et al., 2000, Nielsen, 2000): persuasive evidence argues for the distinctiveness of sleep stages in general. Although the various major sleep stages share features in common, can oscillate back and forth unpredictably, and may hybridize and give rise to not-easily-classified transitional stages (Nielsen, 2014), meaningful statements can nonetheless be made about their characteristic patterns of phenomenology, electrophysiology, and neurochemistry (Hobson et al., 2000, Nielsen, 2000). In the following sections we present these multiple lines of evidence in support of the utility and plausibility of distinctive (if not entirely mutually exclusive) sleep stages. We argue that these neurophysiological and phenomenological idiosyncrasies lead to the strong and testable hypothesis that patterns of brain activation, as measured by relatively non-invasive functional neuroimaging methods like fMRI and positron 	   44	  emission tomography (PET), should vary accordingly across the NREM and REM sleep stages. Moreover, the finding that REM sleep (with high chances of dreaming) shares many neural correlates with waking self-generated thought (Domhoff, 2011, Fox et al., 2013, Domhoff and Fox, 2015), coupled with the knowledge that self-generated thought frequency and vividness differ markedly across sleep stages (Hobson et al., 2000, Nielsen, 2000), leads to the even stronger hypothesis that specific brain areas involved in self-generated thought should show corresponding activation and deactivation throughout the sleep cycle in concert with subjectively-experienced differences in content reports. The body of this chapter provides a systematic and quantitative meta-analytic treatment of all neuroimaging studies of sleep and dreaming in humans, in an effort to synthesize what has been learned from two decades of investigations of brain activations and deactivations throughout the sleep cycle. The aim is not to argue for strict one-to-one isomorphisms between  mental states (ψ) and neuromarkers (Φ), but rather to summarize the current evidence for broad and intriguing correspondences between first- and third-person levels of description. 	  Phenomenological	  dissociation	  of	  sleep	  stages	  	   The subjectively-experienced (‘phenomenological’ for our purposes) differences in the experience of sleep stages have been noted for millennia (Aristoteles and Gallop, 1996) (Deutsch and Dalvi, 2004, Sharma, 2012, Thompson, 2014), but systematic research using large, representative samples has taken place mostly in the past few decades (Nielsen, 2000, Domhoff, 2003). Although many methodological difficulties (and almost as many theoretical deadlocks) have burdened this otherwise burgeoning field, some general conclusions can be cautiously 	   45	  drawn regarding differences in the frequency, quality, and content of mentation across sleep stages.   Nielsen (1999; 2000) has thoroughly summarized this literature, highlighting the critical distinction between cognitive activity in general (which can include thought-like mentation, isolated flashes of imagery, and so on) and truly immersive and hallucinatory dreaming proper (a particular subset of cognitive activity). Whereas cognitive activity is fairly prevalent throughout all sleep stages (at least 40% of awakenings from any given sleep stage will lead to a report of some kind of cognitive activity; Nielsen, 1999), dreaming proper is largely restricted to REM sleep and certain sub-stages of NREM1 sleep onset, and is comparatively rare during other NREM sleep stages (Table 4).   Table	  4.	  Frequency of subjective reports of cognitive activity and dreaming across sleep stages.	  Sleep Stage Cognitive Activity Dreaming Predominant EEG rhythm NREM1 (sleep onset) ~40% ~35% Alpha, Theta NREM2 50% ~15-25% Theta, Spindles (Beta) NREM3/4 (SWS) 40-50% ~10% Delta REM 80% 80% Theta, Beta LREM 100% 100% Theta, Beta, Gamma EEG: electroencephalography; LREM: lucid REM; NREM: non-rapid eye movement; REM: rapid eye movement; SO: sleep onset; SWS: slow-wave sleep. Table based upon comprehensive reviews by Nielsen (1999; 2000) and Hobson et al. (2000).  	  Electrophysiological	  dissociation	  of	  sleep	  stages	  	   The first and best-known neurophysiological (Φ) division of sleep is based on scalp electrode recordings of pooled neuronal electrical potentials, presumed to represent primarily the summation of post-synaptic potentials throughout dendritic arbors and cell somata, and to a 	   46	  lesser extent, synchronous discharge (action potential firing) of populations of neurons (Olejniczak, 2006, Buzsáki et al., 2012).  The central findings regarding EEG correlates of sleep are summarized in Table 4 and Figure 4. Briefly, resting (eyes-closed) wakefulness is characterized by alpha rhythms (8-12 Hz); the transition to NREM1 (sleep onset) is defined by the gradual disappearance of alpha and the appearance of theta (4-7 Hz) ripples and rolling eye movements. NREM2 begins when high-frequency spindles (in the beta frequency; 12.5-30 Hz) and large-amplitude K-complexes appear frequently in the EEG. NREM3/4 or SWS is characterized instead by very slow delta band (0.5-4 Hz) activity, synchronized in large-amplitude waves. Finally, REM sleep involves a return to highly desynchronized and low voltage activity predominantly in the theta and beta bands, similar to active wakefulness. Lucid REM sleep, in the few investigations so far conducted, involves an EEG pattern similar to REM sleep but with increased power in the gamma (>30 Hz) band (Voss et al., 2009, Voss et al., 2014). As these electrophysiological differences are almost universally accepted, well-validated, and expertly reviewed elsewhere (Williams et al., 1974, Antrobus, 1991, Silber et al., 2007), they are not discussed further here.  	   47	  	  Figure	  4.	  Main electrophysiological correlates of each sleep stage. All NREM1 sub-stages are included, as is eyes-closed waking rest, for comparison. Note the easily-differentiable EEG signature accompanying each sleep stage. Reproduced from Stenstrom, Fox, et al. (2012). 	  	   48	  Neurochemical	  dissociation	  of	  sleep	  stages	  	   The patterns of neurochemical activity throughout the sleep-wake cycle are exceedingly complex and consequently poorly understood. At least a dozen major neurotransmitters are involved in regulating NREM and REM, by virtue of either increased or decreased activity (compared to waking) during various sleep stages, but many secondary players with less clear roles are also involved (Kahn et al., 1997, Gottesmann, 1999, Hobson et al., 2000, Pace-Schott and Hobson, 2002, Lena et al., 2005, Stenberg, 2007). Moreover, changes in neurotransmitter levels are far from a global phenomenon throughout the brain: region-by-transmitter interactions have in some cases been experimentally demonstrated (e.g., for dopamine: (Lena et al., 2005)), and it seems likely that activity levels for other neurotransmitters will also vary based not just on sleep stage, but also which region of the brain is being investigated. These complexities are further exacerbated by the existence of many receptor subtypes for each neurotransmitter, and the concomitant (and often unknown – at least for sleep) differences in downstream effects caused by the actions of a single neurotransmitter (Monti and Monti, 2007). A final and major difficulty is that implanting recording electrodes directly into the subcortical nuclei responsible for manufacturing and/or disseminating these neurotransmitters is the ideal method of investigation – an approach typically precluded in humans. Almost everything that is known about the neurochemistry of sleep has therefore been drawn from studies of animals with variable phylogenetic proximity to humans, such as rats, cats, rabbits, and monkeys (Jones, 1991, Gottesmann, 1999, Jones, 2005, Lena et al., 2005, Stenberg, 2007).  The enormous difficulty of studying the neurochemistry of sleep and dreaming have so far precluded the formulation of a clear model of each neurotransmitter’s relative activity throughout NREM and REM sleep stages, much less what the functional implications of such neurochemical heterogeneity might be. Nonetheless, decades of research have yielded some 	   49	  broad trends that appear to be generally reliable, and are bolstered by converging evidence across species, brain regions, and methods. We summarize these general trends in Table 5.  Although a stage-by-stage model is premature given the current limits of our knowledge, broad trends easily distinguish waking at least from REM and NREM sleep. Generally speaking, all major neurotransmitters show some level of tonic activity during waking; conversely, all of these neurotransmitters show a greater or lesser decrease of activity during various NREM stages of sleep (it should be noted, however, that much of this data is derived only from the later stages of ‘slow-wave’ NREM sleep). Finally, REM sleep shows an intermediate pattern: most neurotransmitter activity is decreased relative to waking, but notably, acetylcholine and dopamine levels are elevated (Table 5).   	  	  	  	  	   	  	   50	  Table	  5.	  Neurochemical profiles of the various stages of sleep as compared to waking.	  Neurotransmitter State REM (↑  Dreaming) NREM (↓  Dreaming) ACh ↑↑ ↓ Asp ↓ ↓↓ DA ↑↑ ↓ GABA ? ? Glu ↓↓ ↓ HA ↓↓ ↓ NE ↓↓ ↓ 5-HT ↓↓ ↓ ACh: acetylcholine; Asp: aspartate; DA: dopamine: GABA: 𝛾-amino butyric acid; Glu: glutamate; HA: histamine; NE: norepinephrine/noradrenaline; NREM: non-rapid-eye-movement sleep; REM: rapid-eye-movement sleep; 5-HT: serotonin; SWS: slow wave sleep. Based on data from (Gottesmann, 1999, Pace-Schott and Hobson, 2002, Lena et al., 2005, Hobson, 2009). 	  	  Functional	  neuroimaging	  of	  sleep	  and	  dreaming:	  a	  fourth	  line	  of	  evidence	  	   The preceding sections have highlighted some of the key evidence supporting the hypothesis that stages of sleep are readily dissociable in terms of their (i) phenomenology, (ii) electrophysiology, and more provisionally, (iii) neurochemistry. These many heterogeneities across sleep stages lead to the strong and testable hypothesis that brain activations and deactivations, as detected with relatively non-invasive functional neuroimaging modalities, might also show dissociable patterns throughout the sleep cycle. Functional neuroimaging therefore provides a fourth line of evidence that could either corroborate (or possibly contradict) the ample body of research suggesting phenomenological and neurophysiological dissociability. As noted above, the phenomenological variations in self-generated thought across sleep stages further suggest that differential activation patterns, if they exist, should exhibit some relationship 	   51	  with the numerous regions implicated in self-generated thought in waking (Fox et al., 2015) and REM sleep (Fox et al., 2013). 	  Aims	  of	  the	  present	  meta-­‐analysis	  	   Despite a profusion of neuroimaging studies elucidating the neural basis of sleep and dreaming (Pace-Schott, 2007, Maquet, 2010, Desseilles et al., 2011), no meta-analysis providing a comprehensive overview has yet been conducted. The increasing number and wide diversity of reported brain activations and deactivations observed across different sleep stages, however, demands a meta-analytic synthesis. The drastically different subjective content reported from laboratory awakenings across the various sleep stages corroborates the distinctive patterns of observed brain activity, but there has yet to be any systematic attempt to relate these differing first-person experiences to particular patterns of brain activation. Recently, we conducted such a meta-analysis and review of subjective content for REM sleep and dreaming (Fox et al., 2013); here, we expand upon those results to include functional neuroimaging and subjective report data from all sleep stages. Beyond being the first quantitative meta-analysis of functional neuroimaging studies of sleep and dreaming, this is the first review to attempt a systematic relating of subjective experience, electrophysiology, neurochemistry, and brain blood-flow-related activity across each sleep stage. 	   	  	   52	  Table	  6.	  Brain states investigated, with number of contributing studies and foci, and number of meta-analytic results obtained.	  Cognitive / Brain State Contributing Studies (N) Contributing Foci (A/D) Hypnagogia / NREM1  4 27/47 Light Sleep / NREM2  4 17/32 Deep Sleep / NREM3/4 (SWS) 7 30/150 Dreaming / REM 8 84/49 Lucid dreaming / LREM 1 10/0 A: activations; D: deactivations; LREM: lucid REM; NREM: non-REM; REM: rapid eye movement; SWS: slow wave sleep. 	  	  Methods	  Study	  selection	  Three experimenters scoured Google Scholar, PubMed, and PsycInfo online databases for any study whose title or abstract included keywords such as ‘PET,’ ‘fMRI,’ and ‘neuroimaging,’ in combination with keywords such as ‘sleep,’ ‘dreaming,’ ‘NREM,’ or ‘REM.’ The reference list of each study found was also consulted, as were the bibliographies of several major review of functional neuroimaging of sleep (Hobson et al., 2000) Our search yielded 58 functional neuroimaging (PET or fMRI) studies of REM and NREM sleep. As inclusion criteria, we required that studies employ a baseline of resting wakefulness (either pre- or post-sleep), in order to minimize the confounding effects of various tasks and baseline conditions; that studies involve neurological and psychiatrically healthy participants; and that studies reported peak coordinates for contrasts of interest in standard stereotactic space (either Talairach or MNI coordinates). A total of 16 studies met these inclusion 	   53	  criteria (Table 7). Because many studies examined more than one sleep stage, a total of 24 contrasts of interest were obtained from this pool of studies.  Besides failing to meet our basic inclusion criteria, reasons for exclusion included the addition of extraneous factors (e.g., auditory stimulation during sleep), inclusion of clinical populations or pharmacological agents, failure to provide information for peak foci of activation, or lack of an appropriate baseline (e.g., studies comparing REM sleep during phasic rapid-eye-movement events with regular tonic REM sleep). Note that many studies employed a sleep deprivation paradigm to aid participants in sleeping in the scanner (see discussion of limitations, below). 	  Table	  7.	  Studies included in meta-analyses of sleep and dreaming.	  Study Modality N Stage(s) investigated Sleep deprivation? Maquet et al. (1996) PET 11 REM Y Braun et al. (1997) PET 37 SWS, REM Y Maquet et al. (1997) PET 11 SWS Y Nofzinger et al. (1997) PET 6 REM N Braun et al. (1998) PET 10 REM Y Kajimura et al. (1999) PET 18 SWS Y Finelli et al. (2000) PET 8 REM Y Maquet et al. (2000) PET 5 REM N Peigneux et al. (2001) PET 12 REM N Balkin et al. (2002) PET 27 NREM2 Y Kjaer et al. (2002) PET 8 NREM1 N Maquet et al. (2005) PET 22 SWS, REM N Kaufmann et al. (2006) fMRI 9 NREM1, NREM2, SWS Y Picchioni et al. (2008) fMRI 4 NREM1 N Andrade et al. (2011) fMRI 25 NREM1, NREM2, SWS N Koike et al. (2011) fMRI 12 NREM2, SWS, REM N 	  	   54	  Activation	  likelihood	  estimation	  (ALE)	  meta-­‐analysis	  We used a quantitative, random-effects meta-analytic method known as activation likelihood estimation (ALE) (Turkeltaub et al., 2002, Laird et al., 2005, Eickhoff et al., 2009, Eickhoff et al., 2012) implemented in the software program GingerALE 2.3.1 (San Antonio, TX: UT Health Science Center Research Imaging Institute). The most recent ALE algorithm tests for above-chance clustering of peak foci from different experiments included in the meta-analysis (Eickhoff et al., 2009, Eickhoff et al., 2012) by comparing actual activation foci locations/clustering with a null distribution created by distributing the same number of foci randomly throughout the brain, through several thousand iterations. Included activation foci were smoothed using a full-width half maximum (FWHM) Gaussian kernel dependent on the sample size (subjects) of the experiment from which foci were drawn (larger sample -> smaller smoothing kernel – empirically determined by (Eickhoff et al., 2009, Eickhoff et al., 2012)). Resulting statistical maps show clusters where convergence between activation foci is greater than would be expected by chance (i.e., if foci from each experiment were distributed independently). Statistical maps were thresholded using a false discovery rate (FDR – (Genovese et al., 2002)) of q = .10 (pN) and a cluster threshold of k = 100 mm3. We used these relatively liberal height and extent thresholds due to the very small number of studies available for meta-analysis. To display results, we used template brain images from GingerALE 2.3.1 displayed in the ‘Mango’ software package (San Antonio, TX: UT Health Science Center Research Imaging Institute). Final figures were created using MRIcron. 	  	   55	  Reporting	  of	  meta-­‐analytic	  results	  	   Though we recently conducted a similar meta-analysis of neuroimaging studies of REM sleep (Fox et al., 2013), we update and expand that analysis here. We conducted a total of 8 separate meta-analyses: meta-analytic assessments of both whole-brain activations and deactivations for each of four sleep stages (NREM1, NREM2, NREM3/4 [SWS], and REM). In addition, the results of a single fMRI investigation of lucid dreaming (Dresler et al., 2012) were consulted for comparison purposes. Activation and deactivation maps were the concatenated onto a single brain template for visualization purposes. 	  Results	  Neural	  correlates	  of	  NREM1	  sleep	  	   For NREM1 sleep, the most salient finding was widespread visual network activations, including clusters in the fusiform gyrus bilaterally (BA 20/36/37), the middle occipital gyrus (BA 19), and the cuneus (BA 18). Deactivations were most prominent in the thalamus bilaterally, in somatomotor cortex (BA 4), in premotor cortex bilaterally (BA 6), and in right dorsolateral prefrontal cortex (BA 9).   	   	  	   56	  Table	  8.	  Activations and deactivations associated with NREM1 sleep.	  Region Cluster Size (mm3) Side Peak Coordinates  (x, y, z) Peak ALE value Activations Fusiform gyrus (BA 20/36) 152 R 44, -38, -24 0.0089 Fusiform gyrus (BA 37) 152 L -22, -44, -18 0.0089 Internal capsule 152 L -26, -24, 0 0.0076 Superior temporal gyrus (BA 22) 152 L -52, -28, 2 0.0089 Middle occipital gyrus (BA 19) 152 R 42, -68, 8 0.0089 Cuneus (BA 18) 152 R 16, -82, 30 0.0089 Premotor cortex (BA 6) 152 R 28, 10, 48 0.0076 Precuneus (BA 7) 112 L -18, -66, 40 0.0073 Deactivations Thalamus 1240 R 4, -20, 6 0.0116 Somatomotor cortex (BA 4) 448 L -45, -8, 52 0.0081 Premotor cortex (BA 6) 320 L -36, 6, 40 0.0073 Thalamus 280 L -6, -14, 16 0.0067 Premotor cortex (BA 6) 184 R 14, 8, 56 0.0078 Dorsolateral prefrontal cortex (BA 9) 160 R 31, 23, 28 0.0071 Inferior parietal lobule (BA 39/7) 160 L -31, -56, 44 0.0073 Precuneus (BA 7) 160 L -14, -63, 46 0.0073   Neural	  correlates	  of	  NREM2	  sleep	  NREM2 was associated with marginal results overall, consistent with the very small sample of studies included in this meta-analysis. Activations included some default network (BA 32) and visual network (BA 18) regions, as well as clusters in the cerebellum and caudate nucleus. Deactivations were present in the inferior parietal lobule bilaterally (BA 40), dorsal anterior cingulate cortex (BA 24), and the cerebellum, among other regions (see Table 9).  	   	  	   57	  Table	  9.	  Activations and deactivations associated with NREM2 sleep.	  Region Cluster Size (mm3) Side Peak Coordinates  (x, y, z) Peak ALE value Activations Cerebellum 160 M -2, -77, -28 0.0087 Caudate nucleus (body) 160 R 12, 6, 9 0.0087 Anterior cingulate cortex (BA 32) 160 M -8, 28, 19 0.0087 Lingual gyrus (BA 18) 152 M 8, -88, -8 0.0089 Superior temporal gyrus  (BA 21/22) 152 L -64, -8, -4 0.0089 Posterior mid-cingulate cortex (BA 24) 152 M 6, -12, 36 0.0089 Somatomotor cortex (BA 4) 136 L -46, -6, 56 0.0089 Superior temporal gyrus (BA 42) 128 R 68, -18, 8 0.0089 Deactivations Cerebellum 152 M -6, -68, -34 0.0071 Inferior temporal gyrus (BA 20/37) 152 L -56, -62, -12 0.0071 Inferior parietal lobule (BA 40) 152 R 46, -30, 48 0.0071 Precuneus (BA 7) 152 R 14, -50, 60 0.0071 Premotor cortex (BA 6) 152 R 28, 6, 62 0.0071 Inferior parietal lobule (BA 40) 144 L -48, -28, 60 0.0071 Dorsal anterior cingulate cortex (BA 24) 128 M -8, 5, 46 0.0058       	   58	  Neural	  correlates	  of	  NREM3/4	  sleep	  (SWS)	  SWS was associated with modest clusters of activation in the superior temporal gyrus bilaterally (BA 22), in right parahippocampal cortex (BA 19), and in various visual network regions (BA 18/19). In contrast to these small clusters of activation, massive deactivations were observed throughout the brain, most notably in the thalamus, and in prefrontal executive regions, including orbitofrontal cortex bilaterally (BA 47), right rostrolateral prefrontal cortex (BA 10), left ventrolateral prefrontal cortex (BA 47), and dorsal anterior cingulate cortex (BA 32). Further, deactivations were observed in default network regions, including posterior cingulate cortex (BA 31) and multiple clusters in medial prefrontal cortex (BA 32).  	   	  	   59	  Table	  10.	  Activations and deactivations associated with NREM3/4 sleep (SWS).	  Region Cluster Size (mm3) Side Peak Coordinates  (x, y, z) Peak ALE value Activations Superior temporal gyrus (BA 22) 256 L -56, -23, 2 0.0077 Cerebellum 160 M 2, -58, -18 0.0081 Parahippocampal cortex (BA 19) 160 R 20, -59, -13 0.0079 Middle temporal gyrus (BA 37) 160 L -46, -49, -8 0.0081 Superior temporal gyrus (BA 22) 160 L -46, -34, -2 0.0082 Lingual gyrus (BA 18) 160 M 10, -85, 0 0.0081 Posterior insula (BA 13) 160 L -40, -23, 2 0.0079 Middle occipital gyrus (BA 19) 160 L -36, -78, 10 0.0079 Lingual gyrus (BA 18) 152 M 10, -60, 6 0.0084 Deactivations Thalamus 2336 L -6, -22, 2 0.0180 Thalamus/hypothalamus/subgenual anterior cingulate cortex (BA 25) 1312 M 0, -6, -6 0.0150 Lateral globus pallidus/putamen 776 R 16, 8, -6 0.0137 Orbitofrontal cortex (BA 47) 536 L -20, 20, -22 0.0147 Caudate nucleus (body) 528 R 14, 6, 8 0.0141 Orbitofrontal cortex (BA 47) 504 R 20, 24, -18 0.0133 Rostrolateral prefrontal cortex (BA 10) 504 R 26, 52, -12 0.0127 Medial prefrontal cortex (BA 32) 504 M 2, 40, 10 0.0134 Anterior insula (BA 13) 488 R 44, 16, -10 0.0116 Dorsal anterior cingulate cortex (BA 32) 464 M -4, 22, 30 0.0125 Dorsal pons 456 M -6, -34, -36 0.0117 Mid-insula (BA 13) 456 R 38, 6, 2 0.0121 Posterior cingulate cortex (BA 31) 432 M -2, -34, 34 0.0120 Medial prefrontal cortex (BA 32) 392 M 10, 48, -12 0.0116 Lateral globus pallidus/putamen 296 L -24, -4, -6 0.0099 Caudate nucleus (body) 176 L -16, 4, 10 0.0087 Ventrolateral prefrontal cortex (BA 47) 104 L -32, 32, -18 0.0091 Rostrolateral prefrontal cortex (BA 10) 104 R 42, 52, -6 0.0091   	   60	  Neural	  correlates	  of	  REM	  sleep	   REM sleep was associated with recruitment of widespread brain regions. Most prominent were large clusters of activation throughout the medial temporal lobe bilaterally; multiple regions within the default mode network, including two clusters in medial prefrontal cortex (BA 24 and 9/32), dorsomedial prefrontal cortex (BA 9), and orbitofrontal cortex (BA 25); and numerous visual network areas, centered on the lingual gyrus (BA 18/19). Deactivations were most salient in prefrontal executive regions, including bilateral dorsolateral prefrontal cortex (BA 9), bilateral ventrolateral prefrontal cortex (BA 47), and bilateral frontopolar cortex (BA 10). See Table 11 for full results.  	   	  	   61	  Table	  11.	  Activations and deactivations associated with REM sleep.	  Region Cluster Size (mm3) Side Peak Coordinates  (x, y, z) Peak ALE value Activations Parahippocampal cortex (BA 36) 1120 R 24, -42, -12 0.0118 Caudate nucleus (body) 856 M -6, 16, 12 0.0137 Pons/midbrain 848 M 8, -12, -18 0.0125 Parahippocampal cortex/lingual gyrus (BA 36/19) 832 L -18, -50, -12 0.0133 Lingual gyrus (BA 19) 800 R 28, -68, 0 0.0149 Medial prefrontal cortex (BA 24) 704 M 4, 32, 4 0.0138 Paracentral lobule (BA 41) 640 M 0, -14, 46 0.0107 Parahippocampal cortex/lingual gyrus (BA 19) 608 R 22, -60, -10 0.0125 Parahippocampal cortex  (BA 28/35) 600 R 22, -18, -16 0.0126 Pons 464 M -2, -18, -30 0.0108 Parahippocampal cortex (BA 35) 344 L -16, -26, -24 0.0094 Dorsomedial prefrontal cortex  (BA 9) 216 M -2, 44, 28 0.0096 Lingual gyrus (BA 18) 208 L -10, -94, -12 0.0083 Orbitofrontal cortex (BA 25) 184 M -2, 20, -22 0.0093 Medial prefrontal cortex (BA 9/32) 144 R 12, 44, 10 0.0078 Deactivations Dorsolateral prefrontal cortex (BA 9) 648 R 32, 44, 24 0.0161 Ventrolateral prefrontal cortex (BA 47) 592 R 38, 38, -12 0.0141 Posterior cingulate cortex (BA 31)/ Splenium 520 L -10, -34, 26 0.0111 Ventrolateral prefrontal cortex (BA 47) 504 L -30, 40, -12 0.0114 Frontopolar cortex (BA 10) 440 R 18, 50, -16 0.0111 Inferior parietal lobule (BA 39) 368 R 56, -56, 24 0.0101 Inferior parietal lobule (BA 39) 192 L -38, -50, 28 0.0095 Posterior corona radiata 160 R 28, -44, 20 0.0092 Posterior corona radiata 160 R 30, -52, 24 0.0089 Dorsolateral prefrontal cortex  (BA 9) 160 R 44, 32, 24 0.0092 	   62	  Region Cluster Size (mm3) Side Peak Coordinates  (x, y, z) Peak ALE value Inferior parietal lobule (BA 40) 160 R 46, -46, 24 0.0092 Orbitofrontal cortex (BA 11) 152 R 22, 32, -26 0.0086 Rostrolateral prefrontal cortex (BA 10) 152 L -32, 54, -14 0.0086 Rostrolateral prefrontal cortex (BA 10) 152 L -36, 56, 0 0.0086 Dorsolateral prefrontal cortex  (BA 46) 152 L -50, 40, 2 0.0086 Insula/inferior frontal gyrus (BA 13/44) 152 R 46, 18, 10 0.0095 Posterior corona radiata 152 R 28, -32, 20 0.0096   	   	  	   63	  	  Figure	  5.	  Brain activations and deactivations throughout the sleep cycle.	  Brain	  activations	  and	  deactivations	  throughout	  the	  normal	  sleep	  cycle	  (panels	  a-­‐l),	  and	  in	  lucid	  REM	  sleep	  dreaming	  (panels	  m-­‐o).	  Frontoparietal	  control	  network	  regions	  are	  deactivated	  throughout	  all	  normal	  sleep	  stages,	  except	  lucid	  dreaming,	  consistent	  with	  the	  lack	  of	  cognitive	  control	  and	  meta-­‐awareness	  in	  these	  sleep	  stages.	  Default	  and	  visual	  network	  areas,	  in	  contrast,	  show	  a	  stepwise	  relationship	  with	  sleep	  stage:	  they	  are	  most	  activated	  when	  rates	  of	  dreaming	  and	  other	  cognitive	  activity	  are	  highest	  (REM	  sleep	  and	  lucid	  REM),	  and	  deactivated	  when	  phenomenological	  content	  approaches	  zero	  (as	  in	  NREM3/4).	  Lucid	  REM	  sleep	  figure	  reproduced	  from	  Dresler	  et	  al.	  (2012).	  aIns:	  anterior	  insula;	  aIPL:	  anterior	  inferior	  parietal	  lobule;	  dACC:	  dorsal	  anterior	  cingulate	  cortex;	  DLPFC:	  dorsolateral	  prefrontal	  cortex;	  Hth:	  hypothalamus;	  lOcc:	  lateral	  occipital	  cortex;	  LTC:	  lateral	  temporal	  cortex;	  mOcc:	  medial	  occipital	  cortex;	  mPFC:	  medial	  prefrontal	  cortex;	  PCC:P	  posterior	  cingulate	  cortex;	  RLPFC:	  rostrolateral	  prefrontal	  cortex;	  Thal:	  thalamus.	  	   64	  Discussion	  As we observed stepwise relationships between subjective thought content and brain activation throughout the sleep cycle, we discuss overall trends in our results as a whole rather than focusing the discussion on one sleep stage at a time, as in the Results section. Note, however, that the putative links between subjective experiences and brain activation discussed here are almost entirely inferential and based on statistical averages of content reports generated in studies that did not use PET scans to assess brain activation. That is, as outlined in the introduction, the subjective experiences assumed to hold, on average, for each sleep stage are based on datasets entirely independent from those upon which brain activation data are based. The apparent linkages we speculate on here are therefore, at best, crude approximations of average experiential intensity and average brain activation across many different subjects; although intriguing, these correspondences need to be further explored and corroborated but much detailed research before they can be considered reliable, much less definitive. This major concern is discussed in more detail in the subsequent section on Limitations. 	  Default	  and	  visual	  network	  activation	  and	  their	  relationship	  to	  self-­‐generated	  thought	  and	  imagery	  throughout	  the	  sleep	  cycle	  Dreaming can be thought of as an unconstrained, hyper-associative, and highly immersive form of self-generated thought that is largely detached from external sensory inputs (Windt, 2010, Fox et al., 2013, Christoff et al., in press). We therefore expected that areas involved in memory recall and recombination, self-referential thinking, and audiovisual imagery would show heightened recruitment, compared to a restful waking baseline, in all sleep stages that have some appreciable amount of thought or dream content (especially REM sleep, but also potentially stages NREM1 and NREM2). 	   65	  Generally speaking, the NREM sleep stages are associated with considerably lower rates of dreaming than REM sleep (see Introduction) – and the mentation that does occur tends to be less visuospatial and immersive, and more of a conceptually-focused inner monologue (Hobson et al., 2000, Nielsen, 2000, Nir and Tononi, 2010). During NREM1, however, highly vivid visual imagery and occasionally full-blown (if short-lived) dreaming can occur (Nielsen, 1992, Hori et al., 1994, Hayashi et al., 1999, Stenstrom et al., 2012). Consistent with these phenomenological reports, our meta-analysis showed that NREM1 was primarily associated with widespread activation of secondary and tertiary visual regions, including the fusiform gyri bilaterally, the middle occipital gyrus (BA 19), and the cuneus (BA 18). We also observed activation in primary auditory cortex (superior temporal gyrus), consistent with the high rates of auditory imagery experienced at sleep onset (Stenstrom et al., 2012). Absent, however, were any notable activations in default network regions; this absence of activation, however, can be reconciled with the typical brevity and diminished sense self (compared to other sleep stage mentation reports) that characterize sleep onset imagery (Cicogna et al., 1998). Late in the night, when REM sleep predominates in the sleep cycle, NREM2 can give rise to a high frequency of immersive dream experiences indistinguishable from REM sleep reports (Cicogna et al., 1998). Cognitive activity of some kind is relatively frequent during early-night NREM2 (~50% of awakenings), but dreaming proper is more rare, reported from roughly 15-25% of NREM2 awakenings (Goodenough et al., 1965, Nielsen, 1999, 2000). Consistent with these intermediate levels of self-generated cognition and dreaming, NREM2 shows recruitment of numerous default network regions key to self-generated thought, including medial prefrontal cortex (BA 32), lateral temporal cortex bilaterally, and posterior cingulate cortex. Visual network regions also showed modest activation, particularly in the lingual gyrus (BA 18).  	   66	  The frequency of dream experience is lowest in NREM3/4 (SWS) (Hobson et al., 2000, Nielsen, 2000). Indeed, significant numbers of participants can never recall cognitive activity or dreaming of any kind from SWS awakenings, despite multiple nights spent in the laboratory (Cavallero et al., 1992). Consistent with this modest level of self-generated thought, SWS showed some modest clusters of activation in various default network regions, including medial temporal lobe and left lateral temporal cortex, as well as small activations in visual network areas, including the middle occipital gyrus and lingual gyrus. Tempering these modest activations, however, were much larger and statistically stronger clusters of deactivation throughout major default network hubs, including medial prefrontal cortex (BA 32) and posterior cingulate cortex (BA 31). Further, deactivations were apparent in multiple subcortical brain areas, including the hypothalamus, thalamus, and dorsal pons. These subcortical deactivations are consistent with the overall decreased arousal and blockade of sensory inputs in SWS (Hobson et al., 2000). Consistent with our predictions and prior work (Fox et al., 2013), REM, the sleep stage with by far the highest rates of dreaming (80-90% of the time (Hobson et al., 2000), showed heightened activation in numerous regions implicated in self-generated thought and imagery, especially widespread activation of the medial prefrontal cortex, the medial temporal lobe, and medial occipital areas. All of these activations are consistent with the self-generation of a self-referential narrative situated in a largely visual imaginal world. These overall trends in activation patterns across the sleep stages are paralleled by changes in functional connectivity: connectivity among default mode network hubs, for instance, decreases monotonically throughout the NREM sleep stages (Sämann et al., 2011, Wilson et al., 2015). Consistent with these results, PET investigations have found a monotonic decrease in 	   67	  cerebral energy metabolism across NREM stages 1-3 (Maquet, 1995), whereas energy metabolism in REM sleep is equal to (Maquet et al., 1990, Madsen et al., 1991, Braun et al., 1997) or higher than (Heiss et al., 1985, Buchsbaum et al., 1989) waking rest. Finally, lucid REM sleep (LREM), in stark contrast to NREM sleep, shows many activations greater even than non-lucid REM sleep. The most striking difference is the reappearance of activity in the frontoparietal control network, discussed in more detail in the following section. Also apparent is heightened activity in areas already hyperactive in REM sleep, including medial prefrontal cortex and a large swathe of medial occipitoparietal cortex – potentially explained by anecdotal reports that lucid REM sleep experiences are much more vivid and detailed than regular REM dreams (Green, 1968, Sergio, 1988, Yuschak, 2006, Dresler et al., 2012). 	  	  Brain	  deactivation,	  cognitive	  control,	  and	  meta-­‐awareness	  throughout	  the	  sleep	  cycle	   As discussed above (Chapter 2), waking self-generated thought involves a co-activation of default network areas alongside executive frontoparietal control network regions, most notably dorsal anterior cingulate cortex, rostrolateral prefrontal cortex, and anterior inferior parietal lobule. These results are not particularly difficult to rationalize when it is recalled that cognitive control and meta-awareness, the principal functions tied to the latter network, are in fact quite prevalent in waking self-generated thought (even if lower than in typical externally-directed cognition and tasks; recall the discussion in Chapter 1 on cognitive control and metacognition in waking self-generated thought).   Conversely, executive and metacognitive functioning is largely absent or deficient in NREM and REM sleep cognition. Although dreams reports show strong thematic continuity with 	   68	  the emotional and personal concerns of waking life (Cartwright et al., 1984, Kuiken et al., 2008, Fox et al., 2013), nonetheless actual goal-related thinking or top-down control and planning are rare. Further, activities involving sustained top-down control of attention, such as reading, writing, or using a phone or computer, occur only very rarely in dreams (Hartmann, 1996, Schredl, 2000). Although logical and paralogical thinking indeed take place in sleep and dreaming respects (Hall and Van de Castle, 1966, Kahn and Hobson, 2005), overall such thinking is only peripherally goal-related at best and is deficient in many other respects (Kahn and Hobson, 2005).  Metacognitive functioning is similarly compromised. Natural rates of meta-awareness of one’s true state during sleep (i.e., lucid dreaming or lucid sleep) are probably only about 1% even in experienced lucid dreamers (Snyder and Gackenbach, 1988, Schredl and Erlacher, 2004, 2011), and might occur only a handful of times throughout the entire lifespan in normal individuals (Barrett, 1991, Zadra et al., 1992). Moreover, employing a variety of interventions in an effort to increase meta-awareness during sleep (including psychological training, pharmacological agents, and external electrical stimulation), even among highly-motivated participants, tends to result in only very modest and poorly-validated gains (Stumbrys et al., 2012). Meta-awareness of other features of experience, such as bizarre or impossible situations and discontinuities of time and place (Dorus et al., 1971), is likewise heavily compromised (Kahn and Hobson, 2005).  Consistent with these many executive and metacognitive deficiencies throughout the sleep cycle, all sleep stages we investigated showed evidence for deactivation of frontoparietal control network regions. NREM1 showed deactivation of dorsolateral prefrontal cortex (BA 9) as well as premotor cortices (BA 6) bilaterally. In NREM2, dorsal anterior cingulate cortex (BA 	   69	  24) and premotor cortex (BA 6) were similarly deactivated. SWS showed widespread deactivation of the executive system, including in orbitofrontal cortex bilaterally (BA 47), right rostrolateral prefrontal cortex (BA 10), dorsal anterior cingulate cortex (BA 32), and left ventrolateral prefrontal cortex (BA 47). Finally, REM sleep similarly showed large clusters of deactivation in the dorsolateral prefrontal cortex (BA 9/46) bilaterally, rostrolateral prefrontal cortex (BA 10) bilaterally, and ventrolateral prefrontal cortex (BA 47), also bilaterally. In stark contrast to all of these results throughout the rest of the sleep cycle, the unusual state of lucid REM (LREM) sleep instead showed activation of frontoparietal control network regions, including rostrolateral and dorsolateral prefrontal cortices, as well as in the anterior inferior parietal lobule bilaterally – consistent with restored cognitive control and meta-awareness (Dresler et al., 2012, Fox and Christoff, 2014). Overall, these findings can be summarized in a tentative model of how subjective experience and sense of control/meta-awareness co-vary throughout the sleep cycle with activation in various brain regions that fall mostly with the default, visual, and frontoparietal control networks (Fig. 6). The coupling between default and visual network activation and the intensity of cognitive activity/dreaming on the one hand, and the orthogonal but equally intriguing coupling of frontoparietal control network activity with level of cognitive control and meta-awareness, is readily apparent (Fig. 6). 	  	   70	  	  Figure	  6.	  A speculative model of brain-mind isomorphisms across the sleep cycle. Brain activation and subjective experiences in waking self-generated thought (second point from left) are taken as an arbitrary baseline or zero-point, in relation to which brain activation and subjective experiences in other states of sleep and waking are compared. The y-axis employs arbitrary units for illustrative purposes only, to show relative brain activation/deactivation or increases/decreases in subjective intensity in each sleep stage, relative to waking self-generated thought. All values are therefore relative to an origin point defined by waking self-generated thought. Data are drawn from the meta-analyses detailed in the results section above, and from the review of content reports across sleep stages described in the introductory section. Some trends are apparente: for instance, DMN and VN activity broadly track subjective reports of the frequency and vividness of self-generated experience for each sleep stage: sleep stages with greater and more vivid subjective experiences and imagery tend to have greater activation in DMN and VN areas; conversely, SWS, in which subjective experiences and imagery are extremely rare, shows deactivations in DMN regions. A similar correspondence is shown between levels of cognitive control and meta-awareness and FPCN activation: FPCN activation is high, for instance, in externally-directed tasks where cognitive control is corresondingly high, but both cognitive control and FPCN activation fall in most sleep stages; however, some limited recovery of FPCN activation and cognitive control take place in so-called ‘lucid’ dreaming (far right point), in which the subject attains control and metacognitive awareness of the dream state. For further details, see the results and discussion section, above. DMN: default mode network; FPCN: frontoparietal control network; SWS: slow wave sleep; VN: visual network. 	  Limitations	  	  Four major limitations of our results and speculative model should be emphasized. The first is that the subjective reports of nighttime dreaming and cognition have been collected 	   71	  largely independently of the functional neuroimaging data that speaks to brain recruitment during the various sleep stages. That is, although differences in subjective experience across sleep stages are in general reliable and well-replicated (Nielsen, 1999, Hobson et al., 2000, Nielsen, 2000), the studies that have used functional neuroimaging to examine these same sleep stages have rarely actually collected dream or mentation reports from their participants (for instance, of the 8 studies we meta-analyzed for REM sleep, only one confirmed dreaming had indeed been taking place in the REM sleep periods examined in the PET scanner: (Maquet et al., 1996)). In the absence of such reports collected directly following functional neuroimaging of sleep and dreaming, any putative relationship between sleep stage neurophysiology and subjective content remains, at best, inferential and probabilistic. The obvious solution to this problem is for future functional neuroimaging studies of sleep to actively awaken and interrogate participants as to their subjective experiences across various sleep stages. A second major concern is the small number of studies (k) in our meta-analytic sample. Although we took pains to search the literature thoroughly and include every well-controlled and rigorously executed study, nonetheless the field of neuroimaging of sleep remains small. Accumulating research in this domain, however, will gradually mitigate this concern, as more powerful and representative syntheses become possible. Third is the fact that most studies sleep deprived their participants the night before brain scanning in order to facilitate the maximum amount of sleep in the scanner. Aside from nonspecific effects, such as stress, sleep deprivation is known to affect the architecture and EEG correlates of the sleep cycle (Borbély et al., 1981), and might therefore influence neuroimaging measures of brain recruitment as well. In principle this concern could be addressed by comparing brain activation and deactivation for given sleep stages across studies that did and did not 	   72	  employ sleep deprivation, but a much larger body of research is required before any such comparison is possible.  A fourth and final concern that should be reiterated is that sleep stages are as much a convenient abstraction as they are a concrete neurophysiological fact. We refer the reader back to the introductory material for a more detailed consideration of this important issue.  Conclusions	  In a major review more than 15 years ago, Hobson et al. (2000) defended the position that REM and NREM sleep stages “can be defined, that their components can be analyzed and measured, and that they are significantly different from one another” (original emphasis). We have here provided substantial new evidence in favor of this view from a comprehensive meta-analysis of many functional neuroimaging studies across every stage of the sleep cycle. We have presented preliminary evidence that subjective experiences and neurophysiological markers co-vary across the sleep cycle, with the most intriguing finding being that DMN and visual network activation tracks the occurrence and immersiveness of self-generated thought, whereas frontoparietal control network deactivation tracks the general loss of cognitive control and meta-awareness of one’s psychological state. Overall, these results provide an intriguing example of complex, but nonetheless coherent, patterns of brain-mind isomorphism (Cacioppo and Tassinary, 1990). Our neuroimaging findings and accompanying theoretical account will hopefully serve as a useful stepping-stone on the path to a much deeper understanding of sleep and dream neurophysiology that surmounts the major limitations inherent in present research. 	   	  	   73	  CHAPTER	  4	  –	  STUDY	  3:	  SPECIFIC	  NEURAL	  RECRUITMENT	  ASSOCIATED	  WITH	  DIFFERENT	  TYPES	  OF	  THOUGHT	    Introduction	   Chapter 2 presented a detailed meta-analysis of general brain recruitment associated with self-generated forms of thought, but could offer only speculations about the specific functional roles played by the various regions and networks consistently recruited. Major questions remain, however, about the specific neural recruitment associated with specific types of thought – questions that cannot be addressed, beyond simple reverse inference, without targeted experiments.  Investigating	  abstract	  qualities	  of	  thought	   In the Introduction (Chapter 1), we reviewed the bewildering variety of first-person content that characterizes self-generated thought. This multiplicity of specific thought content can be simplified to some extent by examining higher-level, more abstract qualities of self-generated thought. Rather than referring to specific content, abstract qualities of thought involve the high-level functionality or category to which the thought can be ascribed. Some major examples of abstract qualities of thought under which virtually an infinite variety of specific contents can be categorized are summarized in Table 12. Aside from assessing the relative frequency of each of these thought qualities in a healthy adult sample, another aim was to assess their co-dependence (if any). In principle each of these major qualities should be orthogonal to the others – goal-related thinking should be able to arise spontaneously or be intentionally directed, for instance, and ought to be able to feel either pleasant or unpleasant, depending on the 	   74	  specific thought content. Nonetheless, there remains the possibility that thought qualities might co-vary.  The first major quality we sought to investigate is the relation of thought to the immediate environment, or its stimulus-dependence. Here, we use stimulus in the general sense outline in the Introduction (Chapter 1), to encompass any sensory input from the external environment, with external operationalized in its literal sense as beyond the boundary of the body. Stimulus-dependent thought, then, is any thought directly related to or triggered by incoming visual, auditory, tactile, olfactory, or gustatory sensation. Many studies, including the present investigation (see Methods, below), deliberately minimize sources of such external input in order to increase the relative frequency of stimulus-independent thought – the object of interest. Stimulus-independent thought, then, is any other form of cognitive-affective activity not directly related to or triggered by the surrounding ‘here and now.’ Our aim in this study was to discard trials in which subjects reported experience stimulus-dependent thought about their environments, and instead to resolve different types of stimulus-independent thought with follow-up questions in the fMRI scanner. The first major quality of stimulus-independent thought we investigated was agency: i.e., the subjective assessment of whether one’s thought is being intentionally and volitionally directed/guided, or arises and unfolds in a spontaneous fashion. Note that here we use the term ‘spontaneous’ in the limited sense of not being consciously directed or controlled – elsewhere we have developed a technical definition of spontaneous thought that is beyond the scope of the present investigation (Christoff et al., in press).  A third central quality is the self-assessed utility of thought, i.e. whether thought is judged to be related to one’s current goals and concerns or not. The rationale for addressing this 	   75	  question draws on a large body of psychological research dating back to the early days of mind-wandering research, which collectively suggests that goal- and concern-related thinking plays a large, perhaps dominant, role in self-generated thought (Klinger, 1987, Klinger and Cox, 2004, 2011, McMillan et al., 2013). The prevalence of goal- and concern-related thinking, present in upwards of half of all thoughts, has also spawned a conceptual model of mind-wandering and related forms of thinking, the Current Concerns Theory, which posits that such goal-related thinking is a central feature, and adaptive quality, of self-generated thought (Klinger, 1987, Klinger and Cox, 2004, 2011, McMillan et al., 2013).  Finally, we investigated the valence of self-generated thought (from unpleasant, to neutral, to pleasant), in line with extensive research showing that a wide variety of affect accompanies self-generated thought of various kinds (Fox et al., 2014, Fox et al., in preparation).  Table	  12.	  Major qualities of self-generated thought uninvestigated at the neural level.	  Quality Major Thought Alternatives Overview Stimulus-dependence Stimulus-independent Thoughts not caused by, or not directly related to, the surrounding environment Stimulus-dependent Thoughts caused by, or directly related to, the surrounding environment Agency Spontaneously Unfolding Thoughts unfolding without one’s direct control or guidance Intentionally Directed Thoughts unfolding with one’s direct control or guidance Utility Goal-related Thoughts related in some way to one’s goals or current concerns Goal-unrelated Thoughts unrelated one’s goals or current concerns Valence Negative Unpleasant, negatively-valenced thoughts Neutral Thoughts without an obvious affective component or flat affect Positive Pleasant, positively-valenced thoughts  	   76	  Previous	  efforts	  to	  understand	  specific	  neural	  recruitment	  associated	  with	  different	  types	  of	  thought	   Only a very limited number of studies have addressed brain recruitment associated with specific thought content or abstract qualities, for reasons that will become apparent in the Methods section. Briefly, the central problem is the low signal-to-noise ratio of the fMRI BOLD signal, which necessitates a large number of trials in order for event-related designs to yield robust results. Coupled with the inability to experimentally control the frequency, much less the specific content, of thought, this produces a situation in which it is very difficult to collect an adequate number of trials for multiple types of thought within a reasonable scanning time.  Some preliminary work along these lines has been reported, however. A recent study by Vanhaudenhuyse and colleagues (2011) investigated the involvement of three key default network regions in the ‘intensity’ of internal (vs. external, i.e. perceptual) awareness. They defined internal awareness as “all environmental stimuli-independent thoughts (e.g., inner speech, autobiographical memories, or wandering thoughts).” During continuous fMRI acquisition, subjects were prompted by quasi-random experience-sampling probes that asked them to rate the contents of their awareness on a 4-point scale from strongly external (‘1’) or moderately external (‘2’) through to moderately internal (‘3’) and strongly internal (‘4’). Investigating three key default network regions-of-interest (the medial prefrontal cortex, posterior cingulate cortex, and bilateral medial temporal lobe), they found that activity in these regions correlated positively with the self-reported intensity of internal awareness (Vanhaudenhuyse et al., 2011). This study presents some intriguing preliminary evidence that specific neural correlates accompany thought that is specifically stimulus-independent, but the region-of-interest approach limits its generalizability and reliability.  Another study has investigated different types of affect in self-generated thought. Tusche 	   77	  and colleagues (2014) investigated the emotional content of unconstrained thoughts during task-free rest periods using multivariate pattern analysis. They found that patterns of activity within medial prefrontal cortex predicted the valence (positive vs. negative) of thoughts – not just during the initial session, but also at a follow-up scanning session one week later (Tusche et al., 2014).   To our knowledge, no study has directly investigated neural correlates of intentionally directed/guided thought, or of goal- and concern-related thoughts, and the aforementioned studies investigating stimulus-independence and valence clearly leave many questions unanswered. The present study therefore aimed to improve our understanding of self-generated thought by resolving three major categories of stimulus-independent thought at the neural level.  Methods	  Participants	  	   Participants were 16 healthy young adults (9 female; mean age ±  SD in years = 28.1 ±  4.3). Most participants were graduate students at the University of British Columbia, recruited via snowball (i.e., word-of-mouth referral) sampling. All participants selected for participation in the study were neurologically and psychiatrically healthy; right-handed (to minimize any functional or structural neuroanatomical differences due to language lateralization); and provided written informed consent in accordance with the ethical standards set by the Clinical Research Ethics Board (CREB) of the University of British Columbia, Vancouver, Canada. The study protocol was reviewed and approved by the UBC MRI Research Centre. Participants were compensated $20/hr for their time. 	   78	  Experimental	  design	  	   As discussed at length above (Chapter 2), general neural correlates of self-generated thought have been identified in some two dozen studies, and in at least ten agnostic, whole-brain analyses. These investigations have converged on about a dozen regions that appear to be consistently recruited (Fox et al., 2015), but almost nothing is known about the functional roles of each of these regions except via inference from other investigations that have interrogated a given region’s functionality using task-based designs (Andrews-Hanna et al., 2014). The main hurdle to investigating multiple types of self-generated thought with fMRI is in obtaining adequate statistical power: well-powered event-related fMRI analyses require ~15-20 trials for a given event type, with >25 trials being ideal (Murphy and Garavan, 2005). The problem is that self-generated thought, by definition, cannot be controlled or manipulated by the experimenter; a sufficient number of probes, and consequently time in the scanner, must be allowed to elapse such that a sufficient quantity of each thought type arises spontaneously (Vanhaudenhuyse et al., 2011). Our solution to this problem was to allow for the maximal number of thought probes by (i) repeating thought probes as quickly as possible, provided that they continued to yield high rates of self-reported thought, and did not result in frustration or overstimulation for participants; and (ii) dispensing with any task that would sacrifice scanning time.  In order to (i) maximize the frequency of thought probes without participants feeling overwhelmed, we conducted a pilot behavioral study in which inter-probe intervals (IPIs) of varying length (4, 6, 8, 10, 12, 14, 16, 18, and 20 s) were randomly utilized throughout 5 blocks of 24 probes each. Although rates of self-reported thought were very high (>90%) under all interval lengths, participants nonetheless reported that the 4 and 6 s intervals were too brief and led to confusion and an inability to become immersed in the stream of thought. We therefore 	   79	  used a randomized range of IPIs (8, 10, 12, 14, 16, and 18 s; mean = 13 s), with equal numbers of each IPI, for each block and each participant.  In order to further maximize the scanning time dedicated to self-generated thought of various kinds, we (ii) dispensed with a task altogether. Most prior studies of self-generated thought have employed some simple and boring task paradigm, with the main aims being to induce mind-wandering or other task-unrelated thought, and additionally to provide some objective corroboration of mind-wandering episodes via observable decrements in accuracy or other measures of performance (e.g., reaction time).   As noted in many contemplative traditions, however, thought in no way needs to be ‘induced’ – the unoccupied mind will naturally produce a nearly nonstop stream of thoughts, images, inner speech, and so on (Gunaratana, 2011, Eifring, 2015). Moreover, ‘objective’ measures of attentional lapses do not correlate particularly well (or, often, at all) with actual subjective reports of mind-wandering or being off-task; especially with highly practiced tasks, self-generated thought can co-exist perfectly well with adequate task performance. For instance, in a recent fMRI investigation of mind-wandering from our research group, no significant differences whatsoever were observed in either reaction time or accuracy on a simple sustained attention task when ‘on-task’ vs. ‘off-task’ (i.e., mind-wandering) epochs were compared (Christoff et al., 2009). Moreover, even if a suitable objective behavioral or physiological measure of self-generated thought were known, it would likely reveal nothing about the specific content of thought, which was the object of our investigation here. A task was therefore dispensed with altogether in the present study. In summary, to date no fMRI study has adequately examined multiple types of self-generated thought, because the number of trials required is very large and is uncontrollable when 	   80	  allowing the subject to self-generate thought content; therefore our design allowed for the maximal number of trials (120) within a reasonable scanning timeframe (1.5 hrs, including acquisition of 3D T1 and DTI white matter images for morphometric neuroimaging analysis; see Chapter 6) so that individual thought types could be resolved with sufficient statistical power in the ensuing analyses. Each trial (Fig. 7) involved the participant lying quietly, fixating on a blank black screen for between 8–18s (the ‘thought period’). Thought periods were then randomly interrupted with thought probes. The first probe asked participants whether, just prior to the probe onset, they had been having stimulus-independent, stimulus-dependent, or no thoughts. If the participant responded to either of the latter two options, the trial was terminated and a new ‘thought period’ began. If the participant indicated stimulus-independent thoughts, however, they then proceeded to answer three additional probes about the qualities of these thoughts.  Procedure	  and	  fMRI	  thought	  sampling	  paradigm	  First, all participants reported to our laboratory space at the Brain Research Centre (BRC) inside the UBC Hospital. Upon arrival at the lab space, participants completed a consent form, and were then given a detailed explanation of the study and the thought probe questions by the experimenter. A scripted explanation was used to provide identical examples of each thought type for every participant. Participants were then left alone to engage in two practice blocks (24 probes each) of the study (~15 min), during which they had an unlimited time to classify their thoughts and respond to each probe (Fig. 7). E-Prime 2 was used to present the thought probes (Psychology Software Tools Inc., Sharpsburg, PA). If the participant responded to the first probe that they could recall ‘No thoughts’ or that their thoughts were ‘Stimulus-dependent,’ then that 	   81	  trial ended and a new period of a blank black screen (‘thought period’) began. As noted above, our reason for ignoring stimulus-dependent thoughts was to ensure the maximal amount of scanning time dedicated to stimulus-independent thoughts, instead.   Following the practice blocks, the experimenter discussed the experiment and thought probe questions further with each participant in order to clarify any ambiguities in the questions or in the way in which the individual’s thoughts should be classified. Because each probe had a limited number of discrete responses (indeed, response options were mostly binarized), it was emphasized to participants that they should in every case choose the best answer, acknowledging that in some cases thoughts might fall somewhere in between the two response options provided and might not be easily classifiable. When all questions had been clarified, the participant was escorted to the UBC MRI Research Centre, cleared through screening and security protocols, and entered the MRI with the aid of a staff technician. The exact same thought sampling task as given in the practice blocks was projected onto a screen within the MRI (Fig. 7), and responses were made using an MRI-compatible button-box instead of a standard computer keyboard (as in the practice session). The thought sampling experiment was divided into 5 blocks of 24 probes each, for a total of 120 probes. Each block proceeded automatically, with 8 s allowed for each response to each probe question. At the end of each block the experimenter checked in with each participant as to their physical and mental comfort in the scanner, and allowed participants to take a brief break. Upon completion of the experiment, participants were debriefed, and were compensated $20/hr for their time. Most participants returned within 1-2 days to the BRC lab space to complete an identical behavioral version of the thought sampling paradigm for correlational use with our anatomical scans (for details, see Chapter 6). 	   82	  	  Figure	  7.	  Thought probe questions and sequence (identical for both the behavioral and fMRI versions of the experiment). 	  	   fMRI	  data	  acquisition	  The order of acquisition was first a 3D T1-weighted anatomical scan (~6 min), followed by five T2*-weighted functional runs (~8 min each), and ending with a diffusion tensor imaging (DTI) scan (~8 min). Functional and structural MRI data were collected using a 3.0 Tesla Philips Intera MRI scanner (Best, Netherlands) with a standard head coil. Head movement was restricted using foam padding around the head. T2*-weighted functional images were acquired parallel to the anterior commissure/posterior commissure (AC/PC) line using a single-shot gradient echo-planar sequence (EPI; repetition time [TR] = 2 s, echo time [TE] = 30 ms, flip angle [FA] = 90°, field of view [FOV] = 240 × 240 × 143 mm, matrix size = 80 × 80, SENSE factor = 1.0). The exact number of functional volumes varied for each run for each subject, since the pacing of the thought probe responses was self-timed, but was approximately 6 to 8 minutes (i.e., between 180 and 240 functional volumes per run). A longer scan (12 min) was initiated for each run, so that the participant would not run out of time and have the scan end prematurely; when it was observed that the participant had completed the run in the behavioral software (E-Prime), the scan was prematurely force-terminated. The resulting truncated functional volumes were manually inspected and edited to remove partial volume acquisitions. Each functional volume 	   83	  included 36 interleaved axial slices (3 mm thick with 1 mm skip) covering the entire brain.  Before functional imaging, a 3D T1-weighted structural volume was acquired, and following the functional scans, a DTI image was acquired. Details of these scans are reported in a later chapter detailing our morphometric neuroimaging experiments (Chapter 6).  fMRI	  data	  preprocessing	  and	  analysis	  fMRI data for each subject were preprocessed and analyzed using SPM8 (Statistical Parametric Mapping, Wellcome Department of Imaging Neuroscience, London, UK). Slice timing correction was performed using sinc interpolation and resampling with the middle (18th) slice as a reference point. All functional volumes were realigned to the first volume to correct for between-scan motion. The structural volume was coregistered to the mean functional image and segmented to extract a gray matter image. The segmented structural volume was then spatially normalized to a gray matter image of the Montreal Neurological Institute (MNI) template and resliced to a voxel size of 2 × 2 × 2 mm. The derived spatial transformations were applied to the 20 realigned functional volumes to bring them into standardized MNI space. Finally, the functional volumes were smoothed with an 8-mm full-width at half-maximum (FWHM) isotropic Gaussian kernel to compensate for residual between-subject variability after spatial normalization and to permit application of Gaussian random field theory for corrected statistical inference (Friston et al., 1994). To remove low-frequency drift in the blood oxygen-level dependent (BOLD) signal, the data were high-pass filtered using an upper cut-off period of 128 s. No global scaling was performed. Condition effects at each voxel were estimated according to the general linear model in our whole-brain analyses. The model included (i) the observed time series of intensity values at 	   84	  each voxel, which represented the dependent variable; (ii) covariates modeling session-specific effects (i.e., the six head movement parameters), treated as variables of no interest in our contrasts; and (iii) regressor functions constructed by convolving thought-type-specific stick functions with a synthetic hemodynamic response function. The regressor functions were constructed to model each of the different types of reported thought for each block (i.e., stimulus-independent, stimulus-dependent, no thoughts, spontaneously unfolding, intentionally directed, goal-related, goal-unrelated, and unpleasant, neutral, and pleasant). Valence (affect) was presented as a five-point scale in the actual thought sampling paradigm, but all data was collapsed into three categories (negative, neutral, and positive) to increase power. The onset of the thoughts was set at 8 s prior to the onset of the first thought probe, and modeled for a duration of 8 s. The onset and responses to the thought probes themselves were also modeled and treated as regressors of no interest. As discussed above, data relating to stimulus-dependent thoughts was of no interest in the present study and was therefore also modeled as a regressor of no interest. Group random-effects analyses were then performed for each contrast to compare different types of stimulus-independent thought to one another (i.e., spontaneously unfolding vs. intentionally directed thoughts, goal-related vs. goal-unrelated thoughts, negative vs. neutral + positive thoughts, and positive vs. neutral + negative thoughts). Due to the relatively exploratory nature of the study, the threshold for significance was set at p < 0.001 (uncorrected for multiple comparisons) and extent threshold k > 5 voxels; for visualization purposes in our figures, this threshold was relaxed to p < 0.005 (uncorrected) and k > 5 (Christoff et al., 2009). 	   85	  Results	  Behavioral	  results	  	   Behavioral data for all participants, collapsed across all 5 fMRI runs, is presented in Table 13. As our interest in this study was in the neural correlates particular thought types, rather than individual differences in thought tendencies, this data is not analyzed further here. For more discussion and analysis, however, see Chapters 5 and 6. Table	  13.	  Behavioral data for all fMRI runs.	  Subject % SIT % Spontaneous % Related % (–) % Neutral % (+) Age Sex S03 81.74 82.98 68.09 30.85 23.40 45.74 22 M S05 80.91 80.23 49.40 4.55 65.91 29.55 25 M S06 84.76 70.93 28.57 17.24 52.87 29.89 24 F S07 95.00 81.58 24.11 30.09 33.63 36.28 34 F S08 84.87 96.04 26.73 19.00 52.00 29.00 35 F S09 71.93 100.00 36.25 3.75 65.00 31.25 28 M S10 79.66 76.60 65.59 1.09 43.48 55.43 33 M S11 85.34 50.51 33.33 12.12 51.52 36.36 28 F S12 82.76 68.09 57.45 46.24 33.33 20.43 35 F S13 71.29 87.32 8.70 15.49 70.42 14.08 24 F S14 89.17 60.75 47.66 11.21 72.90 15.89 26 F S15 88.98 100.00 55.24 12.38 33.33 54.29 26 F S17 95.83 98.26 81.74 26.09 13.91 60.00 26 F S18 82.73 60.23 52.75 13.19 32.97 53.85 28 M S19 77.53 66.18 30.43 5.97 65.67 28.36 31 M S31 94.17 62.50 36.28 23.64 26.36 50.00 23 M Mean 84.20 77.64 43.89 17.06 46.04 36.90 28 – SIT: stimulus-independent thought. 	  	  Orthogonality	  of	  thought	  qualities	   To examine whether abstract thought qualities were independent of one another or not, we computed all pairwise correlations, for all thought responses, for all participants (across 120 probe responses). This resulted in three correlation coefficients per subject, for relationships between spontaneity/goal-relatedness; spontaneity/affect; and goal-relatedness/affect. All coefficients were Fisher-transformed (Fisher, 1915) and then averaged to determine the mean 	   86	  relationship between each thought type. As summarized in Table 14, relationships between thought types were generally small and all correlations were non-significant (df = 15, all p’s > .05). With 1-2 outliers (with r’s > .5) removed for each pairwise correlation, these relationships became even smaller, with no more than ~6% of the variance in one thought type explained by any other. Importantly, the weakest relationship was that between directedness/spontaneity of thought and goal-relatedness of thought. This was a critical finding because we anticipated that both intentional direction and goal-relatedness of thought might recruit executive resources, so establishing the orthogonality of these thought types was important to the validity of our subsequent fMRI contrasts.  Table	  14.	  Pairwise correlations between thought qualities.	   Paired Thought Qualities Correlation Coefficient Spontaneity-Relatedness Spontaneity-Affect Relatedness-Affect Mean r –.180 .200 .273 Mean r (with outliers >.5 removed) –.118 .179 .249 R2 (with outliers >.5 removed) .014 (1.4% of variance explained) .032 (3.2% of variance explained) .062 (6.2% of variance explained)   Neural	  correlates	  of	  spontaneity	  and	  directedness	  in	  stimulus-­‐independent	  thought	   Our first contrast involved examining the sense of agency accompanying stimulus-independent thoughts, i.e., whether the thoughts were judged to be spontaneously unfolding or intentionally directed. Thoughts were judged to arise spontaneously (77.6% of the time) much more often than they were deliberately directed (22.4% of the time). The brain contrast for spontaneously unfolding thoughts (> intentionally directed) revealed an array of posterior default network regions that were more highly activated when thoughts unfolded spontaneously (Table 	   87	  15, Fig. 8), including the temporoparietal junction bilaterally, the right parahippocampal cortex, right superior temporal gyrus, and precuneus/posterior cingulate cortex. The right putamen was also significantly activated.  No significant activations were observed for the contrast intentionally directed thoughts > spontaneously unfolding thoughts at the same threshold. At a slightly relaxed threshold (as used for visualization purposes in our figures: t > 2.9467, p < .005, uncorrected, k > 5), a single cluster of activation was observed, localized to the white matter adjacent to the left dorsal anterior cingulate cortex (t = 4.10, Z = 3.31, k = 12, MNI coordinates: -24, 11, 28; Fig. 9).     	  	  	   	  	   88	  Table	  15.	  Activation	  peaks	  for	  spontaneously	  unfolding	  thoughts	  (>	  intentionally	  directed	  thoughts)	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Temporoparietal junction R 40/ 41/ 42/13 57 -40 19 79 5.41 3.97 Temporoparietal junction/ posterior insula L 13/41/ 22 -45 -28 7 88 5.31 3.92 Putamen R – 24 14 -2 9 4.84 3.70 Parahippocampal cortex R 35/36 27 -31 -20 14 4.68 3.62 Superior temporal gyrus R 22/21/ 38 54 -10 -8 6 4.07 3.29 Precuneus/posterior cingulate cortex M 7/31 12 -49 49 14 4.04 3.27 Height threshold t = 3.73 (p < .001, uncorrected), extent threshold k > 5.    	   89	  	  Figure	  8.	  Activations for spontaneously unfolding thoughts > intentionally directed thoughts. IPL: inferior parietal lobule; Ins: insula; LG: lingual gyrus; LTC: lateral temporal cortex; PCC: posterior cingulate cortex; Prec: precuneus; Put: putamen; STG: superior temporal gyrus; TPJ: temporoparietal junction. Height threshold t = 2.95 , p < .005, uncorrected; extent threshold k > 5.     	   90	   	  Figure	  9.	  Sub-threshold activation in white matter adjacent to the dorsal ACC/anterior MCC for intentionally directed > spontaneously unfolding thoughts. Height threshold t = 2.95 , p < .005, uncorrected; extent threshold k > 5.	    	   91	  Neural	  correlates	  of	  goal-­‐relatedness	  in	  stimulus-­‐independent	  thought	   Twelve significant clusters of activation were observed for the contrast goal-related thoughts > goal-unrelated thoughts (Table 16; Fig. 10).  Table	  16.	  Activation peaks for goal-related thoughts (> goal-unrelated thoughts)	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score White matter adjacent to cingulate cortex R  27 -10 31 83 6.65 4.47 Cerebellum R – 15 -40 -41 40 5.88 4.17 Lingual gyrus/fusiform gyrus R 18/19 30 -70 -5 83 4.24 3.39 Thalamus R – 18 -16 1 21 5.52 4.02 Premotor cortex/Mid-cingulate cortex L 6/24/32 -12 8 49 7 5.25 3.89 Lingual gyrus/fusiform gyrus L 18/19 -21 -64 -11 46 5.00 3.78 Mid-cingulate cortex L 31/24 -21 -22 40 21 4.67 3.61 Parahippocampal cortex/ Cerebellum R 36 30 -34 -26 7 4.64 3.60 Cerebellum (lobule VI) L – -18 -34 -26 17 4.52 3.53 Posterior cingulate cortex M 23 9 -58 19 22 4.45 3.50 Retrosplenial cortex M 29/26 -9 -37 13 7 4.14 3.33 Lingual gyrus L 18/19 -9 -64 7 7 3.93 3.21 Height threshold t = 3.73 (p < .001, uncorrected), extent threshold k > 5.  No significant activations were observed for the contrast goal-unrelated thoughts > goal-related thoughts.    	   92	  	  Figure	  10.	  Activations for goal-realted thoughts > goal-unrelated thoughts. aMCC: anterior mid-cingulate cortex; dACC: dorsal anterior cingulate cortex; MCC: mid-cingulate cortex; mOcc: medial occipital cortex; PCC: posterior cingulate cortex; PMC: premotor cortex; Thal: thalamus; VI: cerebellar lobule 6; VIII: cerebellar lobule 8; WM: white matter. Height threshold t = 2.95 , p < .005, uncorrected; extent threshold k > 5.	     	   93	  Neural	  correlates	  of	  affective	  valence	  in	  stimulus-­‐independent	  thought	   Several clusters of activation were observed in the dorsal striatum for the contrast positive > (neutral + negative thoughts), particularly throughout the caudate nucleus. Activations were also observed in the right cerebellum and in the white matter adjacent to the left anterior cingulate cortex, as well as at more liberal thresholds in the orbitofrontal cortex/subgenual anterior cingulate cortex and precuneus (Table 17; Fig. 11).  Table	  17.	  Activation peaks for positively-valenced thoughts (> neutral + negative thoughts)	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Caudate (head) L – -18 29 -8 10 5.54 4.03 Caudate (tail) L – -21 -37 13 22 5.02 3.79 Cerebellum (region) R – 30 -73 -47 34 4.71 3.63 White matter adjacent to anterior cingulate cortex L 32/24 -18 20 28 11 4.49 3.52 Caudate (body) L  -18 -13 19 6 4.20 3.36 Height threshold t = 3.73 (p < .001, uncorrected), extent threshold k > 5.   	   94	  	  Figure	  11.	  Activations for positively-valenced thoughts > neutral and negatively-valenced thoughts. Caud: caudate nucleus; dACC: dorsal anterior cingulate cortex; HPC: hippocampus; PHC: parahippocampal cortex; Prec: precuneus; Thal: thalamus; WM: white matter. Height threshold t = 2.95 , p < .005, uncorrected; extent threshold k > 5.  No significant activations were observed for the contrast negative thoughts > neutral + positive thoughts or for the contrast neutral thoughts > positive + negative thoughts.  	   95	  Discussion	  The preceding results, though preliminary, are the first to resolve several interesting categorical divisions of stimulus-independent and self-generated thought at the neural level. Specifically, we found that the qualities of agency, utility, and valence of stimulus-independent thoughts could all be resolved at the neural level.  Neural	  correlates	  of	  spontaneously	  unfolding	  and	  intentionally	  directed	  thoughts	   For spontaneously unfolding thoughts, we found greater recruitment in a number of posterior default network regions, consistent with the notion that self-generated thoughts can arise without top-down control and capture attention in a bottom-up manner (Ellamil et al., 2016). The most intriguing result was a large cluster of activation in the right medial temporal lobe (parahippocampal cortex), which seems to be critically involved in self-generating thought: many intracranial EEG experiments in humans have demonstrated that stimulation of medial temporal lobe elicits thoughts, memories, and dream-like experiences (see Chapter 8), and activity in medial temporal lobe has been shown to peak just prior to the subjective experience of a spontaneously arising thought or memory, in both intracranial EEG (Gelbard-Sagiv et al., 2008) and fMRI (Ellamil et al., 2016) experiments. Our other results were highly consistent with a recent study we conducted specifically examining the spontaneous arising of thoughts (but not their intentional direction or guidance): similar to this investigation (Ellamil et al., 2016), we also found activations in the right inferior parietal lobule/temporoparietal junction, the posterior insula (albeit in the contralateral hemisphere), and posterior cingulate cortex. Inferior parietal lobule has been repeatedly implicated in bottom-up attentional capture by retrieved memories (Cabeza et al., 2008). The lateral temporal cortex and temporoparietal junction have been 	   96	  hypothesized to be part of a particular default network subsystem involved in mentalizing and conceptual processing during self-generated thought (Andrews-Hanna et al., 2014), potentially therefore supporting spontaneous theory of mind and semantic memory retrieval (Kvavilashvili and Mandler, 2004).  In addition (and in contrast to the study by Ellamil and colleagues), we observed clusters in the lingual gyrus and superior temporal gyrus. Findings in ‘sensory’ cortices such as lingual gyrus, superior temporal gyrus, and posterior insula could be explained, respectively, by visual, auditory, and interoceptive thought and imagery in self-generated thought capturing attention in a bottom-up fashion (see Chapter 1 for a review of thought content detailing the high frequency of all these content types).  For intentionally directed thoughts, we observed only one subthreshold cluster, in the dorsal anterior cingulate cortex/anterior mid-cingulate cortex, a canonical executive control region. The meager results for this analysis are likely due to the small number of intentionally directed thoughts reported by most participants (only ~22% of all stimulus-independent thoughts), despite the long scanning session and large number of thought probes – resulting in weak statistical power. The trend-level involvement of an executive brain region, however, is intriguing and is in line with predictions we have made elsewhere (Christoff et al., 2009, Fox and Christoff, 2014, Fox et al., 2015, Ellamil et al., 2016, Fox et al., 2016). Moreover, these results are bolstered by the finding of greater grey matter concentration in numerous executive regions, including dorsal anterior cingulate cortex, in association with higher rates of intentionally directed thought (see Chapter 6).  	   97	  Neural	  correlates	  of	  goal-­‐related	  stimulus-­‐independent	  thoughts	   A wide variety of activations were observed in relation to goal-related thoughts. Especially prominent were several clusters throughout the mid-cingulate cortex extending into the premotor cortex. The mid-cingulate cortex, particularly its anterior subdivision, has a well-established role in executive control (Vogt et al., 2003) and is a key region of the frontoparietal control network (Spreng et al., 2010, Spreng et al., 2013). Such executive system recruitment is consistent with a role for these regions in planning and simulating solutions to one’s goals (Fox et al., 2015). We also observed numerous clusters in the cerebellum bilaterally. Importantly, the cerebellum is far from functionally homogenous. In addition to somatotopic mapping of motor control, a recent comprehensive meta-analysis has identified an overlapping topography related to the cerebellum’s various roles in cognition and affect (Stoodley and Schmahmann, 2009). Intriguingly, two large clusters we observed in the central portion of the cerebellum (bilaterally) overlap very closely with regions in lobules VI and VIII identified by Stoodley and Schmahmann as implicated in executive functioning (as opposed to working memory, or language, motor, somatosensory, spatial, or emotional processes). Finally, we observed clusters of activation in posterior cingulate cortex, medial visual network regions (e.g., the lingual gyrus), and a sizable cluster in left medial temporal lobe (posterior parahippocampal cortex). These regions might cooperate in creating the simulations of hypothetical scenarios presumably required in thinking about one’s goals and plans for the future (Andrews-Hanna et al., 2014).  Neural	  correlates	  of	  positively-­‐valenced	  thoughts	  Our major results in relation to positively-valenced thoughts were activations throughout the head, body, and tail of the left caudate nucleus. These results are consistent with many neuroimaging studies of positive affect and reward, in which basal ganglia recruitment is the 	   98	  single most consistent finding (Phan et al., 2002). Intriguingly, a recent fMRI case study examined an usual participant who claimed to experience excessive and highly-rewarding daydreaming, and found significantly increased basal ganglia and subgenual anterior cingulate recruitment in this subject during mind-wandering (Mason et al., 2013) – consistent with our results in a normal sample. Our findings in the subgenual anterior cingulate cortex area of medial prefrontal cortex are also congruent with the repeated recruitment of nearby medial prefrontal areas in rewarding emotions (Phan et al., 2002) as well as work that has specifically shown a role for nearby medial prefrontal cortex areas in affect during mind-wandering (Tusche et al., 2014).  We also observed recruitment of the right cerebellum, in the area of lobule VI. As noted above, the cerebellum is far from functionally homogenous, but rather exhibits a specific topography with respect to various cognitive and affective processes (Stoodley and Schmahmann, 2009). Importantly, the cluster of activation we observed in the right cerebellum lobule VI overlaps nearly perfectly with a meta-analytic cluster identified by Stoodley and Schmahmann associated with emotional processing (see their Fig. 1c). As a general observation regarding positively-valenced thought, it should be noted that many of the studies included in the meta-analyses with which our results are compared here included fMRI investigations of experiences that go beyond merely ‘pleasant’ emotion. For instance, some studies also explored neural correlates related to highly-valued stimulis (i.e., ‘valuation’ studies); others involved subjects anticipating or predicting a future reward (i.e., imagining future rewards). While such studies generally involve recruitment of similar brain regions to those seen in the experience of positive affect itself, it should be noted that this neural similarity (but subjective diversity of valuation/reward experience) compounds the difficulty of interpreting our results. 	   99	  We observed no significant results for negatively-valenced thoughts, but this is likely due to the very small number of such thoughts participants reported over the thought sampling session. On the other hand, the lack of any results for neutral > emotional thoughts cannot be attributed to a lack of power, as a very large number of neutral thoughts were reported. A tentative conclusion, then, is that affectively neutral thoughts simply do not lead to any brain recruitment significantly different, or greater, than emotionally-charged thoughts, but future work can hopefully examine this issue more carefully.   Limitations	   Some limitations of the preceding study should be briefly mentioned. First, it appears that, despite our efforts to collect a very large number of thought probes, we still had too few trials to fully resolve every category of thought we examined. Specifically, the low number of negatively-valenced and intentionally-directed thoughts resulted in either marginal or no results whatsoever for the contrasts of interest. Because the experimenter naturally cannot force participants to have higher rates of certain kinds of self-generated thought, the only obvious solution to this problem for future work is to collect more thought probes over a longer scanning session in order to obtain more trials of each thought type and increase the statistical power of subsequent analyses. Another potential solution to this problem is to employ multivariate analysis methods; this possibility is discussed further in the subsequent section on future directions.  Another concern is ecological validity, i.e., whether thoughts reported and experienced in the MRI scanner environment are indeed representative of, and generalizable to, typical self-	   100	  generated thought in everyday life. If not, then the neural correlates of the thoughts that are experienced, too, might differ appreciably. In the subsequent Experiment 4 (Chapter 5), we go to some lengths to address this issue, and therefore refer the reader there for further discussion of this important question.  Another important limitation is that our results speak to the neural basis of specific thought types across individuals, but reveal nothing about the brain basis of interindividual differences in thought content and valence – a major area of interest in relation to clinical disorders of self-generated thought such as depressive rumination (Christoff et al., in press, DuPre and Spreng, in press). In our subsequent Experiments 5 and 6 (Chapter 6) we therefore address precisely this question.  Another major concern is the possibility of prevalent themes of thought content coloring, or even obscuring, the abstract thought qualities we were aiming to investigate here. For instance, many of the participants were graduate students, and therefore many of the thoughts reported as ‘goal-related’ are very likely to have been focused on graduate research, thesis requirements, publication of manuscripts, and so on. Especially among a relatively homogenous group such as our sample, there is the possibility that the same or very similar types of thought content could recur frequently enough to overshadow the actual abstract quality we aimed to investigate. Therefore the possibility cannot be discounted that, for instance, a significant amount of the activations observed for goal-related thoughts could in fact reflect specific thinking and planning related to writing manuscripts, theses, and so on. Two main solutions to this problem are possible: either more detailed thought content reports could be collected, and such commonalities controlled for; or a wider variety of individuals, presumably with highly disparate goals and thought content, could be recruited. The second option, however, would still not be 	   101	  able to control for nearly ubiquitous human thoughts and desires, such as thought about sexual activity. If thought about sexual activity is very common across nearly all individuals, and is correlated (for instance) with positive affect, then the only way to resolve positive affect itself at the neural level, uncolored by common content themes, would be to collect more detailed thought reports and to partial out the effect due to specific thought content. Note, however, that the signal-to-noise ratio of fMRI requires a large number of each type of ‘trial’ – or thought, for our purposes. Allowing such a large number of different types of thoughts to spontaneously arise would likely require inordinate and unrealistic amounts of time spent in the MRI scanner (2-3 hours or more); resolving specific thought content and qualities at this level of detail is therefore likely a project that is at (or perhaps beyond) the boundary of what is technologically feasible with MRI.  Conclusions	  and	  future	  directions	   Our results suggest that different types of self-generated thought can be successfully resolved at the neural level with fMRI using contrasts based solely on first-person reports of experience. Moreover, the activations we observed in relation to each thought type were, broadly speaking, consistent with the putative roles of these brain regions and with prior predictions about the role these areas might play in self-generated thought (Fox and Christoff, 2014, Fox et al., 2015).  The naturally low prevalence of certain thought types suggests that future work will need to collect large numbers of thought events over long scanning sessions in order to improve the discriminability of different thought types and better understand their neural basis, assuming that a general linear model (GLM) approach is used, as in the present analyses. Because of the 	   102	  inherent requirement for thousands of individual contrasts using the GLM method, the intrinsically low signal-to-noise ratio of the fMRI BOLD signal is further attenuated by the need to control for these multiple comparisons in some way, which yet further attenuates the statistical power of generally underpowered GLM analyses. Whereas one solution would therefore be to collect an even larger number of trials to improve statistical power, another potential pathway to greater statistical strength in analyzing the present dataset would be to use multivariate analysis techniques, such as multivariate pattern analysis (Norman et al., 2006, Haxby, 2012). For instance, one research group has used multivariate pattern analysis to decode the emotional valence of mind-wandering episodes (Tusche et al., 2014), and others have been able with some success to predict specific mind-wandering content from brain-wide patterns of intrinsic brain activity as measured with fMRI (Gorgolewski et al., 2014). These methods represent powerful techniques which might help in overcoming the lack of statistical power in certain of our analyses (for instance for intentionally directed and negatively-valenced thoughts). We plan to apply multivariate methods to the current dataset in future analyses intended for publication in peer-reviewed journals. 	   	  	   103	  CHAPTER	  5	  –	  STUDY	  4:	  DOES	  THE	  MRI	  SCANNER	  ENVIRONMENT	  INFLUENCE	  THOUGHT	  CONTENT?	   Introduction	  Before discussing neural correlates of self-generated thought in any further detail, it is worth considering a (to our knowledge) still-unexplored question: does the MRI scanner environment exert a significant influence on thought content? There are good reasons to suspect that the answer to this question is ‘yes.’ Although true adverse events are exceedingly rare (Hawkinson et al., 2012), participants entering the MRI scanner, especially for the first time, often report a wide variety of subjective effects, most notably claustrophobia and anxiety (Huettel et al., 2004). It therefore seems plausible that in a study examining the affective valence and other abstract qualities of self-generated thought, such as the present fMRI investigation (Chapter 4), thought content might be unduly influenced by reactions to the scanner environment. To our knowledge, however, this issue remains uninvestigated.  Here we sought to address this problem in a preliminary way by comparing thought content reported (i) by subjects in the MRI scanner environment with (ii) thought content reported by these same subjects but outside the scanner, seated at a desktop computer in a minimally-stimulating room; and (iii) thought content reported from this same laboratory environment in an independent sample of subjects.     	   104	  Methods	  Participants	  	   Participants were the same as those employed in the fMRI experiment (Chapter 4) and the morphometric neuroimaging experiments (Chapter 6), including a total of 32 unique participants engaged in a variety of conditions (Table 18). 	  Table	  18.	  Participant features for context comparisons. Experimental Setting Sample size (N) Age ± SD (years) Sex MRI Scanner 16 28.1  ± 4.3 9 F, 7 M Laboratory room (all subjects) 32 28.1 ± 4.1 16 F, 16 M Laboratory room (independent of MRI sample) 16 27.8 ± 3.9 7 F, 9 M Laboratory room (same sample as in MRI scanner) 16 28.1  ± 4.3 9 F, 7 M 	  	  Statistical	  analyses	  	   All analyses (t-tests and correlations) were performed in SPSS Statistics 20 (Armonk, NY: IBM Corp.). 	  	  	  	   	  	   105	  Results	  Average	  thought	  content	  across	  contexts	   We found that the average proportion of each type of thought across the various categories we measured was nearly identical across contexts (MRI scanner vs. laboratory environment) (Figs. 12 and 13). The only notable difference was that stimulus-independent thought tended to be somewhat higher in the MRI scanner (84.2% of thoughts) vs. in the laboratory room (75.4% of thoughts). However, this difference was much smaller when the same fMRI participants reported their thoughts in a follow-up behavioral testing session (82.5%), whereas a separate sample that only completed the laboratory testing session had a considerably lower percentage of stimulus-independent thoughts (68.7%). None of the comparisons was significant (all p’s > .05).  	   106	  	  Figure	  12	  Average thought proportions inside and outside the MRI scanner. ALL Lab = thoughts reported by 32 participants in lab sessions at a desktop computer; fMRI Ss in MRI = thoughts reported in the MRI scanner by 16 participants who completed the full fMRI study; fMRI Ss in Lab = thoughts reported by the same fMRI participants in a separate behavioral testing session in the lab space at a desktop computer; Lab ONLY = thoughts averages for the participants who only completed the lab-based testing session and entered the MRI scanner only for brief anatomical scans. Note that ALL Lab = fMRI Ss in Lab + Lab ONLY.	      	   107	  	  Figure	  13	  Radar plot of thought proportions in the MRI scanner and in the laboratory. Details are the same as in the previous figure.  Stability	  of	  thought	  content	  patterns	  at	  the	  individual	  level	  	   We then examined the stability of thought content across contexts at the individual level, i.e. whether individual participants tended to have the same proportion of, e.g., goal-related thoughts in and outside the MRI scanner (Fig. 14). Overall, we found that correlations between thought patterns at the individual level were extremely high across environmental contexts (all |r|(14) ≥  .56, all p’s < .03). Note, too, that many subjects had 1-2 days between testing sessions; our data therefore demonstrate that individual thought content proportions are relatively stable over short time periods, as well. The weakest positive association was with rates of stimulus-independent (vs. stimulus-dependent) thought, although a strong correlation was still observed, r(14) = .59, p = .022. 	   108	  	  Figure	  14	  Stability of thought content patterns at the individual level across the MRI scanner and laboratory environments. Average, group-wide thought content changes very little across contexts (see previous two figures), and this stability is largely true at the individual level as well. Somewhat less reliable is the rate of stimulus-independent thinking (r = .59), but once participants are creating self-generated thought, proportions are highly stable at the individual level. The one exception to these trends is the curious reversal of proportion of positively-valenced thoughts. All relationships were statistically significant: all |r|(14) ≥  .56, all p’s < .03.      	   109	  Discussion	   The	  MRI	  scanner	  environment	  does	  not	  appreciably	  influence	  thought	  content	  Average proportional thought content in and outside of the MRI scanner environment were virtually identical. The only notable difference was a somewhat higher proportion of stimulus-independent thoughts in the MRI scanner as opposed to in the laboratory testing environment. This effect is probably due to the more sensory-restricted environment of the MRI scanner: in addition to general sensory impoverishment, body movement and head motion are severely restricted, preventing bottom-up attentional capture by other sensory stimuli, and although being noisy, the MRI scanner noise is repetitive and devoid of semantic content. Although we went to some lengths to ensure that our laboratory testing room was also minimally stimulating (low lighting, no objects of interest in the testing area, a curtain separating the testing area from the rest of the room, etc.), there are three sensory modalities which likely had a higher rate of meaningful input in the laboratory testing context. (i) Visual: because head movement was permitted in the laboratory room, participants were free to look around the testing area and the few objects it contained, which were still far more numerous than the single visual field of the MRI scanner (staring directly ahead at the head coil cage and reflected computer screen). (ii) Auditory: a variety of auditory stimuli from the adjacent laboratory and hallway spaces in the Brain Research Centre could not be entirely suppressed or controlled for. Although the MRI scanner also produces considerable noise, as noted above this noise is repetitive and devoid of any meaningful content, and therefore probably less likely to trigger stimulus-dependent thoughts, especially after an initial period of habituation. Conversely, overhearing conversations from outside the room seems highly likely to capture thought. (iii) Tactile: whereas participants were lying flat and instructed to move absolutely as little as possible in the MRI scanner, they 	   110	  were free to move their bodies and heads in the laboratory session. Any movement would generate additional sensory feedback (e.g., the feeling of the contact of the chair the participant is sitting in, the chair moving, etc.), all of which could have triggered thoughts that would have been classified as stimulus-dependent by our criteria and instructions (since they originated from tactile inputs arising in the external environment).  Even more important, however, is that the lower rate of stimulus-independent thought in the laboratory sessions appears to be driven largely by an overall lower rate of such thought in our laboratory-only sample of subjects (purple bars). That is, our second sample of participants who only completed the laboratory session appears, by chance, to have experienced lower rates of stimulus-independent thought, and therefore lowered the ‘laboratory’ context average for all participants. These findings suggest that in fact the MRI scanner environment has a negligible effect even on rates of stimulus-independent thought, and that it was instead the heterogeneities across our relatively small groups that mainly drove this difference. In sum, although the MRI environment is even less perceptually-stimulating than the minimalist experimental room in which the behavioral experiment took place, which helps explain the small differences in rates of stimulus-independent thought, overall the MRI scanner environment appears to have a totally negligible impact on thought content – consistent with the idea that patterns of thought are driven largely by stable individual trait-like tendencies. Aside from lending support to the validity of our own fMRI results (Chapter 4), these findings add to the ecological validity of many previous fMRI investigations of self-generated thought (reviewed in Fox et al., 2015). 	  	   111	  Individual	  patterns	  of	  thought	  content	  are	  stable	  over	  time	  and	  across	  contexts	  	   A second finding of our analysis is that individual patterns of thought are highly stable across contexts (MRI scanner vs. laboratory) and short time periods (1-3 days). The first exception to this was the lower correlation between rates of stimulus-independent thought itself, which seems to vary more widely, although the correlation was still considerable (r = .56). The other main exception was positively-valenced thought which, strangely, showed a significant negative correlation (r = -.56)  – in stark contrast to every other thought category, which showed much stronger and positive correlations (all r’s ≥    .84)  across contexts. Despite the exception of positively-valenced thought, overall our results suggest that a ~45 min period consisting of 120 thought probes captures relatively stable individual tendencies in thought content that are largely invariable across contexts and over short time periods. If thought content reported in such a session is indeed representative of trait-like individual differences, then these reports might have validity as individual difference measures that could be associated with individual neuroanatomical heterogeneities in grey and white matter structure – a hypothesis we examine in the following chapter and Experiments.    	   	  	   112	  CHAPTER	  6	  –	  STUDIES	  5	  AND	  6:	  INVESTIGATING	  THE	  RELATIONSHIP	  BETWEEN	  INDIVIDUAL	  DIFFERENCES	  IN	  SELF-­‐GENERATED	  THOUGHT	  AND	  BRAIN	  STRUCTURE	  USING	  MORPHOMETRIC	  NEUROIMAGING	    Introduction:	  the	  relevance	  of	  interindividual	  neuroanatomical	  heterogeneity	  While plasticity in the adult mammalian brain was once thought to be absolutely minimal, if not impossible (Ramón y Cajal, 1928), structural plasticity at the microstructural  scale (dendritic spine density, axon process length, etc.) is now well documented in the fully mature brains of non-mammalian species, rodents, and non-human primates. Although investigation of adult structural brain hetereogeneities remains restricted (due to the limitations of current neuroimaging methodologies) to the mesostructural level in humans (e.g., grey matter volume and concentration, fractional anisotropy, etc.), nonetheless wide evidence is now available for functionally-relevant neuroanatomical heterogeneity in the adult brain (May and Gaser, 2006, Zatorre et al., 2012). Perhaps more impressive still has been the discovery of neurogenesis (growth, differentiation, and proliferation of new neurons) in the adult brain, now observed in song birds (Goldman and Nottebohm, 1983), rodents (Altman and Das, 1965), and a variety of other mammalian species (Fuchs and Gould, 2000) – even humans (Eriksson et al., 1998).  Such changes in brain structure are typically associated with overt physical practice, training, environmental enrichment, and the like, but significant associations have also been found between brain morphology and individual trait variables in both pathological and normal populations. For instance, systematic differences in brain morphology have been reported in large-scale meta-analyses of schizophrenia (Antonova et al., 2004), major depression 	   113	  (Koolschijn et al., 2009), and obsessive-compulsive disorder (Radua and Mataix-Cols, 2009). Morphometric neuroimaging has also revealed neuroanatomical differences associated with less marked differences in personality and behavior, however, such as scores on the ‘Big Five’ personality traits (DeYoung et al., 2010) as well as other measures of personality or cognitive style (Kaasinen et al., 2005, Gardini et al., 2009). Researchers, particularly the pioneering daydreaming researcher Jerome Singer, have long suggested that differences in self-generated thought content might reflect not only the momentary contents of the mind, but differing trait styles of thinking (Singer and Brown, 1960, Singer, 1961, Singer and McCraven, 1961, Singer and McCraven, 1962). Based on results from numerous studies using their Imaginal Process Inventory questionnaire, which assessed many different dimensions of self-generated thought, Singer, Antrobus, Giambra, and their colleagues argued for the existence of three broad styles of thinking: positive-constructive, guilty-dysphoric, and guilty-fear-of-failure (Singer and Antrobus, 1970, 1972, Giambra, 1974, 1978, 1980). Follow-up work also found intuitively appealing associations between these thinking styles and scores on Big Five personality traits: for instance, positive-constructive daydreaming was significantly associated with Openness to Experience, whereas guilty-dysphoric daydreaming was associated with Neuroticism (Zhiyan and Singer, 1997). It therefore seemed plausible to us that, as with other normal and clinical differences in personality, thinking, and behavior, major differences in patterns of self-generated thought might be associated with corresponding differences in grey and white matter structure in the central nervous system. Before describing our own study, we first review the results of a single pioneering study in this domain. 	   114	  Previous	  research	  using	  morphometric	  neuroimaging	  to	  investigate	  self-­‐generated	  thought	   To our knowledge, only a single study to date has investigated aspects of self-generated thought with morphometric neuroimaging (Bernhardt et al., 2014). Bernhardt and colleagues (2014) investigated whether the frequency of task-unrelated thought during either a simple or demanding cognitive task was associated with differences in cortical thickness throughout the brain. During execution of the tasks, participants were periodically given experience sampling probes asking whether their current thoughts were related or unrelated to the task at hand. The authors found that the frequency of task-unrelated thought during the simple, undemanding task predicted increased cortical thickness in a large swathe of medial prefrontal cortex, extending into the anterior and mid-cingulate cortex. An additional large cluster was observed on the left lateral surface, extending from middle frontal regions posteriorly toward the temporoparietal junction. Sub-threshold (trend) clusters were also observed in rostrolateral and dorsolateral prefrontal cortex. These results suggested to the authors a potential neuroanatomical basis of individuals’ tendency to decouple from the here and now and instead engage in self-generated forms of thought, and are consistent with a growing body of fMRI research showing the importance of medial prefrontal regions to self-generated thought (Fox et al., 2015).  Several major limitations of this study encourage caution when interpreting its results, however. The most important caveat is the very small number of thought probes employed (an average of only ~7 probes in each task, for each participant). That is, on the basis of only 7 thought probes administered over a ~7 min task, the authors constructed individual difference scores representing the proportion of task-unrelated thought, which they then used to predict cortical thickness throughout the brain. Clearly, the potential for temporary state effects to influence responses to just 7 thought probes over ~7 min is a serious limitation. The same 	   115	  concern can of course be raised with respect to our own methods, although we employed 120 thought probes over ~45 min, somewhat mitigating these concerns (see the Discussion for detailed arguments on this point). The other major limitation is that only a single dimension of self-generated thought was assessed: its frequency. Although interesting, this study left unexplored the many potentially intriguing individual differences in the particular content and valence of self-generated thought, which we sought to explore in our investigation.  The	  need	  for	  further	  morphometric	  neuroimaging	  of	  self-­‐generated	  thought	   The accumulated evidence that styles of thinking may be individually distinctive and stable, together with the dearth of research examining neuroanatomical correlates of thought content, suggests the need for more research along these lines. Moreover, as noted above, the one study that has conducted such an investigation (Bernhardt et al., 2014), used very few thought probes and investigated only the frequency of task-unrelated thoughts (but not their content). We therefore designed a study to explore neuroanatomical associations with various other dimensions of self-generated thought using a much larger number of thought probes (n = 120). We computed individual difference scores for each of our participants based on their responses to 120 thought probes over a ~1 hr period, and used these scores as predictors of grey matter concentration and fractional anisotropy (fractional anisotropy). Because our thought probes assessed multiple dimensions of self-generated thought, we were able to explore several dimensions simultaneously, including the rate of stimulus-independent thought, the sense of agency or control over thoughts, the tendency of thoughts to be related to one’s goals or current concerns, and thoughts’ affective valence.  	   116	  Methods	  Participants	  	   In addition to the 16 participants who engaged in the fMRI study (Chapter 4), 18 more participants were recruited for the morphometric neuroimaging study, in order to increase sample size to a level capable of yielding reasonably stable correlation coefficients (i.e., N > ~25; (Glass and Hopkins, 1970)). Data collection is ongoing to allow for a much better-powered sample of approximately N = 50 (Yarkoni, 2009). One participant’s entire dataset was corrupted due to an error at the MRI scanner, and another was excluded from analyses due to a severe paucity of stimulus-independent thoughts (n = 12, out of 120 thought probes), which precluded any stable individual measures of particular thought types for this participant. In total, then, 32 participants (16 female; mean age in years ± SD = 28 ± 4.1) completed the behavioral thought-sampling protocol outside of the scanner and had artifact-free 3D T1-weighted anatomical images suitable for voxel-based morphometric analyses of grey matter concentration (Experiment 5). Two further participants had susceptibility artifacts in their DTI scans and were removed from white matter analyses, leaving a total sample for the DTI investigation (Experiment 6) of 30 participants (14 female; mean age in years ± SD = 28 ± 4.0).  Most participants were graduate students at the University of British Columbia, recruited via snowball (i.e., word-of-mouth referral) sampling. All participants selected for participation in the study were neurologically and psychiatrically healthy; right-handed (to minimize any functional or structural neuroanatomical differences due to language lateralization); and provided written informed consent in accordance with the ethical standards set by the Clinical Research Ethics Board (CREB) of the University of British Columbia, Vancouver, Canada. The study 	   117	  protocol was reviewed and approved by the UBC MRI Research Centre. Participants were compensated $20/hr for their time.  Procedure	  and	  behavioral	  thought-­‐sampling	  session	   All participants spent approximately 1 hr completing the behavioral portion of the experiment in laboratory space at the Brain Research Centre (BRC) inside the UBC Hospital. E-Prime 2 was used to present the thought probes (Psychology Software Tools Inc., Sharpsburg, PA). Upon arrival at the lab space, participants completed a consent form, and were then given a detailed explanation of the study and the thought probe questions by the experimenter. A scripted explanation was used to provide identical examples of each thought type for every participant. Participants were then left alone to engage in two practice blocks (24 probes each) of the study (~15 min), during which they had an unlimited time to classify their thoughts and respond to each probe. Following the practice blocks, the experimenter discussed the experiment and thought probe questions further with each participant in order to clarify any ambiguities in the questions or in the way in which the individual’s thoughts should be classified. Because each probe had a limited number of discrete responses (indeed, response options were mostly binarized), it was emphasized to participants that they should in every case choose the best answer, acknowledging that in some cases thoughts might fall somewhere in between the two response options provided and might not be easily classifiable. When all questions had been clarified, the participant was left alone in a minimally-stimulating room with a curtain drawn to complete the full thought-sampling task, seated at a desktop computer. The thought sampling experiment was divided into 5 blocks of 24 probes each, for a total of 120 probes. Each block proceeded automatically, with 8 s allowed for each 	   118	  response to each probe question, but participants had the option of taking brief, self-paced breaks between each block. Upon completion of the experiment, participants recalled the experimenter (who waited outside in an adjacent room), were debriefed, and were compensated $20/hr for their time. Most participants then proceeded directly to the MRI scanner for anatomical scanning (3D T1 and DTI scans); due to some scheduling conflicts, some participants were instead scanned within 1-2 days of their behavioral testing session. 	  Calculation	  of	  individual	  difference	  scores	  in	  self-­‐generated	  thought	   Though responses from each participant were collated into a database and then converted to percentages (proportions) to control for differences in the absolute number of thoughts reported by each participant. This resulted in individual scores for the percentage of stimulus-independent vs. stimulus-dependent thoughts; spontaneously unfolding vs. intentionally directed thoughts; goal-related vs. goal-unrelated thoughts; and positively-valenced, negatively-valenced, or neutral thoughts. All individual scores are reported in the Results section, below. 	  	  MRI	  data	  acquisition	  Imaging was conducted at the UBC MRI Research Centre (Vancouver, British Columbia). MRI data were collected using a 3.0-Tesla Philips Intera MRI scanner (Best, Netherlands) with a standard 8-element 6-channel phased array head coil with parallel imaging capability (SENSE; (Pruessmann et al., 1999)). Head movement was restricted using foam padding around the head. MRI scanning was conducted either on the same day that the participants completed the behavioral training and thought-sampling session, or within 1-2 days if same-day scheduling was not possible. 	   119	  For each subject, we collected (a) a high-resolution 3D T1-weighted anatomical volume (SPGR: TR = 2 s; TE = 3.53 ms; 175 interleaved axial slices covering the whole brain, 1 mm thick with 0 mm skip; FOV: 256 × 256 × 175 mm; matrix size: 256 × 250; 1 x 1 x 1 mm3 isotropic voxels). We also collected (b) a diffusion-weighted anatomical volume using high angular resolution diffusion-weighted imaging (HARDI) (Tuch et al., 2002). The DTI acquisition protocol included a whole-brain sequence with 1 non-diffusion-weighted image and 60 directionally sensitized diffusion-weighted images, with the 61 brain slices oriented obliquely to the anterior commissure–posterior commissure line, 2.2 mm thick with 0 mm skip (TR = 7909 ms, TE = 60 ms, flip angle: 90°, FOV: 224 × 154 × 224 mm, matrix size: 100 × 100, voxel dimensions: 2.5 mm × 2.5 mm × 2.5 mm). All image data were visually inspected for apparent artifacts due to subject motion and instrumental malfunction; two subjects were removed from white matter analyses due to susceptibility artifacts in their DTI data.  Preprocessing	  of	  3D	  T1	  anatomical	  images	  All preprocessing was performed offline using SPM8 statistical parametric mapping software package (Wellcome Department of Imaging Neuroscience, London). Each subject’s high-resolution structural image was segmented into multiple tissue types, including grey matter, white matter, and cerebrospinal fluid. A subject-specific brain template was then created using the DARTEL tool in SPM8  (Ashburner, 2007) to determine the nonlinear deformations required for warping all grey and white matter images to match one another. The subject-specific mean template was then nonlinearly normalized to MNI space. Lastly, an 8-mm full-width at half maximum (FWHM) Gaussian kernel was applied to spatially smooth the data. 	   120	  Preprocessing	  of	  DTI	  images	  Eddy current and head motion correction were done via a linear, affine registration (FLIRT; (Jenkinson and Smith, 2001, Jenkinson et al., 2002, Greve and Fischl, 2009). Non-brain voxels were removed using FSL’s brain extraction tool (Smith, 2002). DTI eigenvalues (where λ1 is along the principal diffusion axis and λ2 and λ3 are orthogonal to λ1) were calculated using software in FSL’s diffusion toolbox (Smith et al., 2004, Jenkinson et al., 2012), from which the relevant DTI parameter (fractional anisotropy) was derived. To identify regions of interest (ROIs), the FA maps of all subjects were registered into the Johns Hopkins University (JHU) 1 mm fractional anisotropy (FA) standard brain template space (Wakana et al., 2007, Hua et al., 2008) via FLIRT (12 degrees of freedom, cost function = correlation ratio) and the nonlinear registration tool FNIRT (Andersson et al., 2007b, a), which uses a b-spline representation of the registration warp field (Rueckert et al., 1999) (Jacobian range = 0.1–10).  Voxel-­‐based	  morphometry	  analysis	  of	  grey	  matter	  concentration	  Voxel-based morphometry (VBM) is a class of techniques for automatically assessing neuroanatomical heterogeneities in grey matter, either between groups or longitudinally in the same subjects (Ashburner and Friston, 2000, Mechelli et al., 2005). Following the preprocessing steps described above, our various individual difference scores were entered into a multiple regression design matrix in SPM8 and correlated with grey matter intensity at each voxel in whole-brain analyses across our sample (N = 32). Because both grey and white matter are known to be negatively impacted by normal aging (Good et al., 2002), age was entered as a covariate of no interest in our analyses (despite the relatively narrow range and low variance of ages in our 	   121	  sample) to control for any effects due to normal age-related atrophy. Sex was also entered as a covariate of no interest.  Tract-­‐based	  analysis	  of	  fractional	  anisotropy	  In order to assess fractional anisotropy, we first defined 20 white matter tract regions-of-interest (ROIs) based on the Johns Hopkins University (JHU) probabilistic white matter tractography atlas (Fig. 15) (Wakana et al., 2007, Hua et al., 2008). This procedure has numerous benefits as opposed to a whole-brain white matter analysis: it excludes all voxel tissue types (e.g., grey matter and cerebrospinal fluid) of no interest to the analyses by creating a white matter tract-exclusive mask, therefore precluding spurious results; it reduces the need for thousands of multiple comparisons at every white matter voxel in the brain, thereby greatly increasing statistical power; and it enhances the interpretability of results by confining findings to major tracts as a whole, since specific functional roles are mostly unknown for particular loci along white matter tracts. Mean FA was calculated for each of the 20 white matter tracts for each participant, and these values were correlated with individual differences scores on thought content. Because of our small sample size (N = 30), and the additional loss of degrees of freedom when controlling for age and sex, we considered this Experiment merely a pilot study and therefore used liberal statistical thresholds. Given the recent criticisms of null hypothesis significance testing and arbitrary p-value thresholds (Cumming, 2012, 2013), we sought to report potentially practically meaningful effect sizes (which are unaffected by sample size) rather than simply searching for statistically ‘significant’ findings (which are highly dependent on sample size). As a criterion for reporting of preliminary results, we therefore set a ‘medium’ effect size 	   122	  (r ≅ .25-.35) as defined by Cohen (Cohen, 1992), i.e., a minimum of r > .25 for our partial correlation coefficients.  	  Figure	  15.	  White matter tract regions-of-interest (ROIs). Mean fractional anisotropy (a standard measure of fractional anisotropy) was calculated for each tract, based on the Johns Hopkins University probabilistic white matter tract atlas (Wakana et al., 2007, Hua et al., 2008), and correlated with individual differences in thought content. ROI image reproduced from Luders et al. (2011).	  	  	   	  	   123	  Behavioral	  results:	  individual	  differences	  in	  thought	  content	  	   All	  thought	  probe	  responses	  were	  condensed	  into	  individual	  percentage	  scores	  for	  each	  thought	  type,	  as	  described	  in	  the	  Methods.	  Results	  are	  reported	  in	  Table	  19.	  	  Table	  19.	  Individual patterns of thought content	  Subject % SIT % Directed % Related % (–) % Neutral % (+) Age Sex S03 81.74 17.02 68.09 30.85 23.40 45.74 22 M S06 79.82 30.23 30.59 12.79 61.63 25.58 24 F S07* 95.00 18.42 24.11 30.09 33.63 36.28 34 F S08 88.14 1.94 36.27 26.21 41.75 32.04 35 F S09 82.35 2.38 8.54 8.43 72.29 19.28 28 M S10 71.17 33.77 84.62 5.33 40.00 54.67 33 M S11 76.99 41.38 45.98 14.94 43.68 41.38 28 F S12 76.79 43.02 58.82 33.73 31.33 34.94 35 F S13 89.29 20.00 22.67 16.00 66.67 17.33 24 F S14 89.19 62.63 54.55 6.06 72.73 21.21 26 F S15 85.71 1.96 50.98 10.78 67.65 21.57 26 F S16 86.67 75.73 79.81 9.62 39.42 50.96 24 F S17 94.92 1.79 88.18 38.39 6.25 55.36 26 F S18 86.24 40.43 59.14 16.30 32.61 51.09 28 M S19 67.02 29.51 28.57 9.68 87.10 3.23 31 M S20 57.39 4.62 11.86 30.00 35.00 35.00 26 M S21 57.76 9.09 20.90 24.24 48.48 27.27 25 F S22* 39.13 6.67 38.64 22.22 28.89 48.89 31 F S24 82.29 27.27 52.11 34.25 26.03 39.73 26 F S25 82.11 28.95 56.52 22.97 52.70 24.32 27 M S26 57.50 37.31 32.35 20.29 43.48 36.23 37 M S27 73.87 21.79 20.78 2.47 34.57 62.96 27 M S28 65.83 24.68 34.62 30.38 60.76 8.86 27 M S29 63.46 33.33 90.91 15.38 32.31 52.31 30 M S30 86.84 44.90 47.47 30.93 43.30 25.77 21 M S31 97.48 50.45 58.62 15.52 39.66 44.83 23 M S32 50.00 14.04 36.84 19.64 50.00 30.36 26 F S33 60.00 37.88 40.00 39.39 21.21 39.39 30 F S34 72.97 39.51 55.00 28.75 17.50 53.75 28 M S35 69.75 13.58 46.34 26.25 32.50 41.25 26 M S36 23.15 20.83 25.00 32.00 28.00 40.00 31 M S37 93.75 2.97 93.33 30.48 32.38 37.14 34 F SIT:	  stimulus-­‐independent	  thought.	  Asterisks	  (*)	  denote	  subjects	  not	  included	  in	  DTI	  analyses	  due	  to	  susceptibility	  artifacts	  in	  diffusion-­‐weighted	  data.	  	  	  	  	  	   	  	   124	  Morphometric	  neuroimaging	  results	  I:	  grey	  matter	  concentration	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  intentionally	  directed	  stimulus-­‐independent	  thought	   We found a large number of clusters of grey matter concentration throughout the brain associated with individual tendency toward intentionally directed thoughts. Most striking were numerous clusters in prefrontal executive regions, including dorsolateral prefrontal cortex bilaterally (BA 46/9/10), bilateral subgenual anterior cingulate cortex/orbitofrontal cortex (BA 25/12), and right ventrolateral prefrontal cortex (BA 47). Also salient were clusters in bilateral posterior parahippocampal/hippocampal cortex (BA 30) and left superior temporal gyrus/anterior insula (BA 22/13). Results are summarized in Table 20 and Fig. 16.  	   	  	   125	  Table	  20.	  Grey matter concentration peaks for higher frequency of intentionally directed thoughts	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Superior temporal gyrus/ insula L 22/13 -50 1 -4 535 5.78 4.65 Subgenual cingulate cortex/orbitofrontal cortex L 25/12 -20 10 -16 575 5.04 4.22 Dorsolateral prefrontal cortex L 46/9 -54 31 20 131 4.86 4.10 Lingual gyrus R 19 21 -59 -1 355 4.62 3.95 Posterior parahippocampal cortex/hippocampus L 30 -30 -33 -4 751 4.54 3.90 Rostrolateral/dorsolateral prefrontal cortex R 10/46 39 39 -1 93 4.30 3.74 Ventrolateral prefrontal cortex R 47 43 22 -9 72 4.22 3.68 Subgenual cingulate cortex/orbitofrontal cortex L 25 -2 3 -13 117 4.19 3.66 Posterior parahippocampal cortex/hippocampus R 30 22 -39 5 81 3.89 3.45 Middle occipital gyrus R 19 39 -78 20 7 3.87 3.43 Superior temporal gyrus R 21 48 0 -15 44 3.82 3.40 Putamen L – -30 -2 8 21 3.80 3.38 Precuneus R 7 20 -44 56 16 3.72 3.33 Premotor cortex R 6 18 -5 64 20 3.71 3.32 Cerebellum L – -36 -33 -51 12 3.67 3.29 Dorsal anterior cingulate cortex/premotor cortex L 32/6 -10 21 42 35 3.65 3.27 Lateral globus pallidus L – -16 0 -4 6 3.52 3.17 Height threshold t = 3.41 (p < .001, uncorrected), extent threshold k > 5.     	  	  	  	   126	  	  Figure	  16.	  Grey matter concentration peaks for higher frequency of intentionally directed thoughts. Height threshold t = 2.76 (p < .005, uncorrected), extent threshold k > 5. aMCC: anterior mid-cingulate cortex; BA: Brodmann area; dACC: dorsal anterior cingulate cortex; DLPFC: dorsolateral prefrontal cortex; Ins: insula; mOcc: medial occipital cortex; MTL: medial temporal lobe; PCC: posterior cingulate cortex; PMC: premotor cortex; Prec: precuneus; sgCC: subgenual cingulate cortex; STG: superior temporal gyrus.	  	  	   	  	   127	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  goal-­‐related	  stimulus-­‐independent	  thought	   For correlations between grey matter concentration and frequency of goal-related thoughts, we found two significant clusters, in the posterior cingulate cortex extending anteriorly into the left thalamus, and in the left parahippocampal cortex (BA 35), extending superiorly into the thalamus. At a slightly relaxed threshold used for visualization purposes (p < .005, uncorrected), we observed other clusters of correlation as well in dorsal anterior cingulate cortex anterior mid-cingulate cortex (BA 24), left dorsolateral/rostrolateral prefrontal cortex (BA 10/9), left inferior parietal lobule (BA 40), and in right parahippocampal cortex (BA 35/36). Results are summarized in Table 21 and Fig. 17.   Table	  21.	  Grey matter concentration peaks for higher frequency of goal-related thoughts	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Posterior cingulate cortex/thalamus M 30/29/ 26 -4 -33 12 266 4.17 3.65 Parahippocampal cortex/thalamus L 35 -16 -32 -4 30 4.00 3.53 Height threshold t = 3.41 (p < .001, uncorrected), extent threshold k > 5.   	   128	  	  Figure	  17.	  Grey matter concentration peaks for greater goal-related thoughts. Height threshold t = 2.76 (p < .005, uncorrected), extent threshold k > 5. aMCC: anterior mid-cingulate cortex; dACC: dorsal anterior cingulate cortex; DLPFC: dorsolateral prefrontal cortex; IPL: inferior parietal lobule; L: left; PCC: posterior cingulate cortex; PHC: parahippocampal cortex; R: right; Thal: thalamus.   	   129	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  positively-­‐valenced	  stimulus-­‐independent	  thought	   We then examined correlations between grey matter concentration and frequency of positively-valenced thoughts. We found significant associations in a large swathe of anterior mid-cingulate cortex (BA 24), in posterior mid-cingulate cortex (BA 24), bilaterally in the inferior parietal lobule (BA 40), and in the right inferior temporal gyrus (BA 20). See Table 22 and Fig. 18.  Table	  22.	  Grey matter concentration peaks for higher frequency of positively-valenced thoughts	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Anterior mid-cingulate cortex M 24 2 10 27 96 4.11 3.60 Inferior parietal lobule L 40 -51 -33 36 21 4.05 3.56 Posterior mid-cingulate cortex M 24 4 -6 30 34 3.86 3.43 Inferior parietal lobule L 40 -46 -33 51 50 3.82 3.40 Inferior temporal gyrus R 20 54 -26 -19 6 3.54 3.19 Height threshold t = 3.41 (p < .001, uncorrected), extent threshold k > 5.  	   130	  	  Figure	  18.	  Grey matter concentration peaks for greater positively-valenced thoughts. Height threshold t = 2.76 (p < .005, uncorrected), extent threshold k > 5. IPL: inferior parietal lobule; ITG: inferior temporal gyrus; LG: lingual gyrus; MCC: mid-cingulate cortex; PHC: parahippocampal cortex.     	   131	  Grey	  matter	  concentration	  associated	  with	  frequency	  of	  negatively-­‐valenced	  stimulus-­‐independent	  thought	   We found several clusters, mostly spread throughout premotor cortices, as well as in the left cerebellum and precuneus. Results are summarized in Table 23 and Fig. 19.   Table	  23.	  Grey matter concentration peaks for higher frequency of negatively-valenced thoughts	     MNI coordinates    Region Side BA x y z Voxels t-statistic Z-score Premotor cortex R 6/32 20 18 44 215 4.95 4.16 Precuneus L 7 -12 -66 51 32 3.94 3.48 Cerebellum M – -9 -71 -25 57 3.78 3.37 Premotor cortex R 6/32 24 18 71 24 3.77 3.36 Premotor cortex R 6 18 7 59 8 3.73 3.33 Premotor cortex M 6 2 24 57 9 3.55 3.20 Premotor cortex M 6 10 13 62 7 3.53 3.18 Height threshold t = 3.41 (p < .001, uncorrected), extent threshold k > 5.   	  Figure	  19.	  Grey matter concentration associated with frequency of negatively-valenced thought. Height threshold t = 2.76 (p < .005, uncorrected), extent threshold k > 5.PMC: premotor cortex; Prec: precuneus. 	   132	  Morphometric	  neuroimaging	  results	  II:	  fractional	  anisotropy	  	   As discussed in the Methods, above, we conducted a preliminary pilot study of the relationship between mean fractional anisotropy (FA) and various individual thought tendencies in white matter tracts throughout the brain. As this was an exploratory analysis, we merely report relationships that showed medium-sized (r = .25) effects or greater. As tendency toward both positive and negative emotion showed essentially identical relationships, we collapsed these into a single category of overall tendency toward having thoughts with emotion (of any valence). All results are summarized in Figure 20. 	  	  	  	   133	  	  Figure	  20.	  Relationship between individual thought tendencies and mean fractional anisotropy (FA) in white matter tracts throughout the brain. Standardized residual values are displayed (Z-scores) for partial correlations controlling for age and sex. A thumbnail image of each white matter tract is displayed for ease of interpretation. CST: corticospinal tract; CgH: cingulum (hippocampal component); IFO: inferior fronto-occipital fasciulus; tSLF: temporal component of the superior longitudinal fasciculus; Unc: uncinate fasciculus. 	   	  	   134	  Discussion	  	  Grey	  matter	  concentration	  and	  intentionally	  directed	  thought	  	   Some of our most intriguing results were observed for those participants who had higher rates of intentionally directed thought. In previous work, we have speculated that executive frontoparietal control network areas recruited during self-generated thought might be specifically involved in directing, guiding, and selecting among thoughts (Christoff et al., 2009, Fox and Christoff, 2014, Fox et al., 2015, Ellamil et al., 2016, Fox et al., 2016), but to date no empirical evidence has addressed this issue. Here, we provide the first direct evidence that the executive system is indeed preferentially tied to the direction and guidance of self-generated thoughts. This evidence is also consistent with our fMRI results (Chapter 4), which found preliminary evidence that dorsal anterior cingulate cortex, a canonical executive control region, is more active during intentionally directed thought. Here, however, we observed more widespread recruitment of executive areas, including dorsal anterior cingulate cortex, left dorsolateral prefrontal cortex (BA 46), right rostrolateral prefrontal cortex (BA 10/46), and both medial and lateral portions of premotor cortex (BA 6).   Grey	  matter	  concentration	  and	  goal-­‐related	  thought	  Individual propensity toward goal-related thought was correlated with increased grey matter concentration in a combination of default and frontoparietal control network areas and subcortical regions. Clusters that fell within the boundaries of the frontoparietal control network included the left dorsolateral prefrontal cortex, dorsal anterior cingulate cortex/anterior mid-cingulate cortex, and a small cluster in the left anterior inferior parietal lobule. Clusters were also 	   135	  observed in default network areas such as the posterior cingulate cortex and right parahippocampal cortex. Finally, significant associations were also found in subcortical regions, including the thalamus and cerebellum. As with intentionally directed thought, we have previously hypothesized that executive brain regions might be preferentially involved in future planning and other forms of thinking about one’s goals and current concerns (Christoff et al., 2009, Fox and Christoff, 2014, Fox et al., 2015, Ellamil et al., 2016, Fox et al., 2016). The present morphometric neuroimaging results, together with our functional neuroimaging results (Chapter 4), provide the first direct evidence that this is indeed the case. Note that grey matter concentration in these regions appears to contribute independently to both the intentional direction and guidance of thought (previous section), and to a higher propensity to have goal-related thoughts. As we showed above (Chapter 4), intentionally directing thoughts does not necessarily lead them to be more goal-related, and considering the very modest correlation (R2 = .014) between the frequency of these two thought types, for practical purposes they can be considered orthogonal. 	  Grey	  matter	  concentration	  and	  the	  valence	  of	  thought	  	   Our results for grey matter concentration associated with thought valence were somewhat puzzling, and generally inconsistent with our fMRI results (Chapter 4). Recall that we also found that rates of positive affect during the behavioral testing session were drastically different in the same subjects in a later fMRI scanning session (see Chapter 5), as opposed to the frequency of all other thought qualities, which was extremely stable within individuals and across time-points and contexts (laboratory vs. MRI scanner). We therefore suspect that our ~1 hr thought sampling session was insufficient to capture ‘trait’ affective styles; our reports more likely reflect transient 	   136	  moods in our participants. We therefore consider it premature to discuss these results further here. 	  White	  matter	  integrity	  and	  self-­‐generated	  thought	  tendencies	  Above, we reported the results of an exploratory pilot study examining relationships between fractional anisotropy (indexed here by mean fractional anisotropy) and individual differences in various thought tendencies. Although this analysis is highly exploratory and data collection is ongoing to improve the power and reliability of our results, a brief discussion is offered here. One intriguing finding is that mean FA is negatively correlated with rate of stimulus-independent thinking in the corticospinal (i.e., pyramidal) tracts bilaterally. This is intriguing because the corticospinal tracts are the major output pathways from the motor cortex to the body’s voluntary musculature, including controlling fine motor skills (Haines, 2004, Krebs, 2011) – i.e. the main pathways mediating voluntary navigation through the world and manipulation of the body within it. Without speculating unduly, it is interesting to note that individuals who think less about their immediate environment, and who instead engage in higher rates of thought decoupled from the here and now, show reduced fractional anisotropy in the major tracts mediating voluntary control of the body and engagement with the external world. We also found that the frequency of emotionally-valenced (as opposed to neutral) thoughts was associated with higher FA in the right, but lower FA in the left, medial temporal lobe (hippocampal) components of the cingulum. The cingulum travels beneath the cingulate gyrus and wraps around at its posterior end to travel through and connect with the medial temporal lobe (Haines, 2004, Krebs, 2011). As such, it is the major pathway allowing for 	   137	  communication between various limbic structures critically involved in emotion, reward, and memory, and has been implicated in affective disorders such as depression as well as disorders characterized by memory impairment, such as Alzheimer’s disease and Mild Cognitive Impairment (Steele et al., 2008, Jones et al., 2013). Although we offer no explanation for the lateralization of these effects, it is intriguing to observe that pathways with an established role in linking limbic brain structures key to emotional processing show anatomical heterogeneities associated with the reported frequency of self-generated emotions.  We also found that the frequency of goal-related thoughts was negatively associated with mean FA in the medial temporal lobe component of the left cingulum, as well as with higher mean FA in the right uncinate fasciculus. The uncinate fasciculus provides major connections between the medial temporal lobe and prefrontal and orbitofrontal cortex (Kier et al., 2004). Consistent with this finding, recall that grey matter concentration was also significantly increased in the right medial temporal lobe (parahippocampal cortex). Finally, frequency of directed thought was inversely correlated with mean FA in the right corticospinal tract and right inferior fronto-occipital fasciculus, as well as directly correlated with mean FA in the temporal component of the right superior longitudinal fasciculus. With respect to the corticospinal tract, and similar to our comments above, it is interesting to observe decreased fractional anisotropy in the brain’s major motor output tract associated with an increased tendency to control and guide thoughts detached from the external environment. The inferior fronto-occipital fasciculus provides the major pathway conveying visual and auditory information to the frontal lobes; again, it is interesting to note decreased fractional anisotropy in a pathway conveying external sensory information coupled to a stronger tendency to focus executive resources on internal channels of information. The superior longitudinal fasciculus 	   138	  represents one of the brain’s major rostrocaudal fiber pathways, linking widespread parietal and prefrontal areas, with putative roles in the top-down control of attention and working memory, among other functions (Makris et al., 2005). Again, though of course speculative, increased fractional anisotropy here is reasonable in individuals who have a greater tendency to direct and guide their internal thought streams. 	  	  Cellular	  foundations	  of	  macroscale	  morphological	  changes	  The underlying cellular-level changes that putatively give rise to the many gross structural differences reported here remain very poorly understood, particularly in humans. Nonetheless, numerous candidate cellular mechanisms have been identified (Anderson, 2011, Zatorre et al., 2012, Lovden et al., 2013) and are tentatively summarized in Table 24. Although most of these relationships remain highly speculative, some first steps have been taken toward understanding cellular-level changes that might give rise to brain morphology differences identifiable via morphometric neuroimaging. Preliminary evidence from a mouse model (Lerch et al., 2011), for instance, found that volumetric changes as measured by morphometric MRI methods correlated most strongly with the presence of axon growth cones (as opposed to neuron size or number). This suggests that axon sprouting and reorganization may be a more important factor than, for example, neuron soma growth or neurogenesis proper. In contrast, neurogenesis – which results in the addition of relatively few total neurons, remains equivocal in humans, and seems mostly restricted to the hippocampus (Zatorre et al., 2012) – is a very unlikely cellular mechanism (for detailed discussion, see the reviews noted above). Determining the cellular candidates for changes observed with respect to individual differences in patterns of thinking faces the additional challenge that direct animal models are not possible, whereas motor skill or 	   139	  spatial learning, for instance, can be plausibly simulated in rodents and monkeys and compared concurrently with human results (Sagi et al., 2012).  Table	  24.	  Candidate cellular mechanisms putatively underlying macroscale structural differences in gray and white matter Cellular	  Mechanism Tissue	  Type	  Affected Morphometric	  Methods	  	  Putatively	  Affected Dendritic	  arborization/	  synaptogenesis	   GM	   GMC,	  GMV	  Neurogenesis	   GM	   CT,	  GMC,	  GMV	  Angiogenesis	   GM	  +	  WM	   CT,	  GMC,	  GMV,	  WMV	  Axon	  sprouting	   GM	  +	  WM	   CT,	  GMC,	  GMV,	  FA	  Gliogenesis	   GM	   CT,	  GMC,	  GMV	  Fiber	  reorganization	   WM	   FA	  Myelin	  formation	   WM	   FA	  Myelin	  remodeling	   WM	   FA	  Astrocyte	  changes	   WM	   WMV	  CG: cortical gyrification; CT: cortical thickness; FA: fractional anisotropy; GM: gray matter; GMC: gray matter concentration; GMV: gray matter volume; WM: white matter; WMV: white matter volume. Based on Anderson (2011), Zatorre et al. (2012), and Lövdén et al. (2013).  	  	  Limitations	   The most significant limitation of these two studies is the uncertainty involved in using thought content measured during a single ~1 hr session as a putative ‘trait’ measure of individual self-generated thought tendencies. Several factors in our design mitigate this concern, however. First, we collected a large number of thought probes per subject (n = 120). To our knowledge, this is far more thought probes than used in any prior neuroimaging study of self-generated thought: Christoff and colleagues (2009), for instance, presented only 16 thought probes per subject; Bernhardt and colleauges (2014), in their morphometric investigation of task-unrelated thought, employed only 7 thought probes (but still used this data as a ‘trait’ measure to 	   140	  predict brain structure differences in cortical thickness). The number of thoughts sampled here was therefore approximately an order of magnitude larger than in previous work.  Moreover, there is evidence that self-generated thought frequency in the laboratory mirrors thoughts experienced in daily life (McVay et al., 2009), and, as we showed above (Chapter 5), overall self-generated thought tendencies appear to be consistent across both the laboratory and MRI scanner environments. Moreover, within the same participants but across several days, we found that thought content is highly similar, both in and outside the MRI scanner (Chapter 5), further supporting the validity of these individual differences in reported content as putative trait measures. More typically, ‘trait’ variables are measured with self-report questionnaires. Although some such questionnaires have been developed to capture qualities of self-generated thought (Singer and Antrobus, 1970, 1972), evidently there are serious limitations to this method of identifying ‘trait’ patterns of thought. Such questionnaires employ retrospective self-assessment, which is potentially limited by distortions of memory (Loftus, 1996) and self-serving bias (Shepperd et al., 2008). Moreover, even though a questionnaire might ask the participant to describe average or ‘trait’ qualities, this is no guarantee that temporary state effects (such as current mood or the day’s events) will not color the participant’s responses – the exact same limitation of using moment-to-moment thought sampling as a putative trait measure.	    	   141	  CHAPTER	  7	  –	  IDENTIFYING	  THE	  NECESSARY	  AND	  SUFFICIENT	  NEURAL	  BASIS	  OF	  SELF-­‐GENERATED	  THOUGHT:	  THE	  VALUE	  OF	  NEUROPSYCHOLOGICAL	  LESION	  STUDIES	   Introduction	  All of the aforementioned results are, at heart, correlational: although we have reported many novel relationships between subjective experience, neurophysiological activity, and neuroanatomical structure, nonetheless these associations cannot demonstrate causal roles for any of these brain regions or networks in self-generated thought. In order to provide strong evidence for a causal role of a given brain region in a given cognitive process or mental state, the gold standard is a true loss-of-function experiment in which the area of interest is ablated or inactivated. Although this kind of research is routinely conducted in animal models, human studies must rely on naturally-occurring or surgically-unavoidable lesions. Meticulous observation and assessment of cognitive-affective deficits following circumscribed brain lesions is perhaps the oldest form of ‘cognitive’ neuroscience, dating back to the pioneering work, for instance, of Paul Broca (Broca, 1861). Widespread investigation of the cognitive-affective sequelae of virtually every conceivable form of brain lesion has since developed into the highly influential field known as ‘neuropsychology.’ Neuropsychology studies the cognitive, affective, and perceptual deficits suffered by patients with various brain lesions. Far from merely cataloguing the effects of rare brain disorders and diseases, however, lesion studies have shed much light on healthy brain functioning, as well as serving as highly effective catalysts for further research (e.g., (Scoville and Milner, 1957, Luria, 1976, Solms, 1997, Gainotti, 2000, Müller and Knight, 2006). Lesion studies can provide important clues about the necessary and sufficient neural basis of a given 	   142	  cognitive process, even a complex one like self-generated thought (Koenigs et al., 2007). To anticipate our conclusions, lesion work suggests that at least four regions play critical roles in various kinds of self-generated thought: medial prefrontal cortex, inferior parietal lobule, medial occipitotemporal cortex, and medial temporal lobe (Fig. 21).  Identifying	  necessary	  and	  sufficient	  neural	  substrates	  of	  self-­‐generated	  thought	  Although we are aware of little neuropsychological work that has directly addressed the effect of brain lesions on self-generated thought, many studies have addressed closely related cognitive processes. Likely the closest parallel is the extensive neuropsychological lesion work undertaken by Solms regarding the necessary and sufficient neural basis of dreaming (Solms, 1997, 2000b). Solms concluded that the areas most critical to dreaming in general are (i) the medial prefrontal cortex and (ii) the temporoparietal junction/inferior parietal lobule. Additionally, a large swathe of (iii) medial occipital cortex, centering on the lingual gyrus, is critical for the visual features of dreaming (Solms, 1997, 2000b, Bischof and Bassetti, 2004). Elsewhere, we have argued at length that the subjective experiences and neurophysiological correlates of dreaming bear a strong resemblance to those of waking mind-wandering and related forms of self-generated thought (Fox et al., 2013, Fox and Christoff, 2014, Domhoff and Fox, 2015). Notably, all three regions identified by Solms as critical to nighttime dreaming emerged as significantly activated in our meta-analysis of waking self-generated thought (Fox et al., 2015; for further discussion see Domhoff & Fox, 2015). Further supporting this view is the finding that, concurrent with the global loss or reduced frequency of dreaming, patients often reported reduced daydreaming and fantasy following the lesions to medial prefrontal cortex and/or the periventricular white matter tracts at the anterior horns of the lateral ventricles and the genu of 	   143	  the corpus callosum (Frank, 1946, 1950, Piehler, 1950, Schindler, 1953). To our knowledge, these latter few investigations are the only studies to have directly assessed some form of waking self-generated thought in relation to brain lesions, and unfortunately the results are largely anecdotal. Nevertheless, overall this convergence across methods (neuroimaging and lesion work) suggests that more rigorous investigations of the quality and content of waking self-generated thought in patients with damage to these three areas could prove informative.  Because self-generated thought so often involves spontaneous memory retrieval (Andrews-Hanna et al., 2010, Fox et al., 2013), lesion work related to semantic and autobiographical memory capacity is also relevant. Solms’ conclusions concerning the importance of the medial prefrontal cortex to self-generated waking and dreaming experience are bolstered by work showing that lesions to medial prefrontal cortex have detrimental effects on both semantic and episodic autobiographical memory retrieval (Philippi et al., 2014). Another potentially critical region, largely disregarded by Solms (1997), is (iv) the medial temporal lobe (Domhoff and Fox, 2015). Bilateral medial temporal lobe lesions are most famously associated with anterograde (and limited retrograde) amnesia (Scoville and Milner, 1957, Milner et al., 1968); less well known are heavy deficits in dreaming that cannot simply be explained by failures of recall. For instance, Korsakoff’s syndrome patients with bilateral medial temporal lobe damage show a marked decrease in dream reports, even when awakened directly from REM sleep in a laboratory setting (Greenberg et al., 1968). Amnesia alone cannot explain these findings, for at least two reasons. First, medial temporal lobe patients still had some dream recall, demonstrating that they still possessed basic recall capacities (e.g., in one study, ~25% of awakenings elicited dream reports (Greenberg et al., 1968), vs. about 80-90% in normal subjects (Hobson et al., 2000). Second, medial temporal lobe patients retain intact working/short-term 	   144	  memory (Milner et al., 1968): anterograde amnesia should not prevent them from reporting upon experiences from just moments ago, immediately upon being awakened in a sleep laboratory setting. The dreams medial temporal lobe patients do report also support this interpretation: “the content of that material which was recalled showed very stereotyped, commonplace features and reflected very little affect” (Greenberg et al., 1968; p. 205; emphasis added). Nor are these findings restricted to Korsakoff’s patients: a study of encephalitis patients with severe medial temporal lobe lesions reports identical findings (Torda, 1969): patients reported far fewer dreams than controls, and reports “were short and simple… [they] contained one scene with recurrent repetition… The dreams lacked imaginative, unusual, or mysterious details and intensive emotions. The content was stereotyped, repetitious…” (p. 280). The reduction in dream frequency with medial temporal lobe lesions is certainly severe enough to warrant further research with respect to the effect of similar lesions on waking self-generated thought. Several studies from recent years provide further corroborative evidence for the importance of the medial temporal lobe in self-generated thought processes. For instance, medial temporal lobe patients are severely impaired in imagining novel fictitious or future events and experiences (Klein et al., 2002, Hassabis et al., 2007, Rosenbaum et al., 2009, Andelman et al., 2010, Kwan et al., 2010, Race et al., 2011, 2013), congruent with the consistent activation of medial temporal lobe during such tasks in healthy subjects (Schacter et al., 2007, Addis et al., 2009, Schacter et al., 2012). Medial temporal lobe lesion patients also suffer deficits in tasks requiring creative and novel patterns of thinking (Duff et al., 2013, Rubin et al., 2014). Although providing only circumstantial support, these findings are intriguing because so much of the content of self-generated thought consists of imagined future scenarios and novel, creative recombination of prior memories into new thoughts and simulated experiences (Klinger, 2008, 	   145	  Andrews-Hanna et al., 2010, Fox et al., 2013). Investigation of the quality and frequency of self-generated thought in medial temporal lobe lesion patients therefore appears to be a promising avenue for future research. 	   146	  	  	   147	  Figure	  21.	  Neuropsychological lesion studies and meta-analysis of fMRI investigations converge on four brain structures that may be critical to self-generated thought. The left column shows brain templates with dark areas indicating regions critical to nighttime self-generated thought (dreaming) as determined by overlapping lesion sites based on CT scans of neurological patients – except panel g, which shows the average site of medial temporal lobe excisions during surgery for intractable epilepsy. In the case of lesions to medial prefrontal cortex (panel a) there is evidence for severe reduction in waking self-generated thought (daydreaming/fantasy) as well. The middle column shows meta-analytic brain activations associated with waking daydreaming/mind-wandering. The right column shows meta-analytic brain activations associated with REM sleep, which is nearly always accompanied by dreaming (r = .8; Hobson et al., 2000). Note that the activation cluster in panel f is somewhat anterior to the temporoparietal junction, approximately in Brodmann area 40. All three approaches converge on four areas: the medial prefrontal cortex (panels a-c); the temporoparietal junction/inferior parietal lobule (panels d-f); the medial temporal lobe (panels g-i); and the medial occipital lobes/lingual gyrus (panels j-l). Panels a and d based on the work of Solms (2000a); panel j based on Solms (1997). Panels b, e, and h based on data from Fox et al. (2015). Panel g based on average medial temporal lobe excisions for severe epilepsy patients as reviewed by Mathern and Miller (2013). Panels c, f, and i based on data from Fox et al. (2013) and Domhoff & Fox (2015). X and Z values represent left-right and vertical coordinates, respectively, in Montreal Neurological Institute stereotactic space. Figure expanded and modified from Domhoff & Fox (2015), with permission.	      	   	  	   148	  CHAPTER	  8	  –	  NEURAL	  ORIGINS	  AND	  ONTOGENY	  OF	  SELF-­‐GENERATED	  THOUGHT	  AS	  REVEALED	  BY	  HUMAN	  INTRACRANIAL	  ELECTROPHYSIOLOGY	  	   	  Introduction	  Neither functional neuroimaging nor lesion studies can answer questions about the detailed, millisecond-scale temporal dynamics of self-generated thought. Two key questions about these temporal dynamics concern the neuroanatomical origins and ontogeny of self-generated thought: where in the brain do self-generated thoughts tend to originate (there may of course be more than one answer), and how does self-generated activity subsequently spread through distributed neuronal networks and give rise to the accompanying subjective experiences of memory recall and novel thought? A third important question centers on the dynamics of interactions between large-scale networks, most importantly the frontoparietal control and default mode networks, which we have discussed in detail elsewhere (Christoff et al., in press). Similar to the lesion studies discussed above (Chapter 7), the ideal method for investigating fine timescale neural activity – intracranial electrophysiology – is only permissible in humans in clinical contexts (most commonly, epilepsy patients). While we therefore did not conduct such an investigation here, much can be gained from a careful review of the many scattered reports of human electrophysiological recording and stimulation experiments over the past century. In this chapter we review the results of more than 100 such studies, expertly synthesized in the comprehensive review of Selimbeyoglu and Parvizi (2010), and summarize what has been learned about brain regions key to the origins (initial generation) and ontogeny (subsequent propagation) of self-generated forms of thought. 	  	   149	  Neural	  origins	  of	  self-­‐generated	  thought	  With respect to sites of origin, there may be more than one central location of generation, or it may be that difficult-to-localize, distributed network activity gives rise to thoughts, and the search for ‘primary’ thought generation regions is misguided. The reality could also be that both mechanisms contribute, depending on the type and content of self-generated thought. Whatever the answers to these quandaries, the evidence for the central role of the medial temporal lobe in thought generation is compelling. The most direct causal evidence comes from the ever-growing body of cognitive studies of patients who have had brain electrodes chronically implanted, or the cortical surface probed with electrical stimulation, for a variety of clinical reasons – usually intractable epilepsy. Since the first studies of this kind nearly one hundred years ago, essentially every brain area has been explored with direct focal electrode stimulation, or passive recording of spontaneous discharges, to a greater or lesser degree. The most relevant finding from this growing body of research is that stimulation of medial temporal lobe structures (i.e., the hippocampus, parahippocampus, entorhinal cortex, and amygdala) very frequently leads to memory recall, immersive thoughts, and hallucinatory, dream-like experiences (Fig. 22). Far from an isolated occurrence, the accumulated evidence supporting this assertion is fairly substantial (Feindel and Penfield, 1954, Bickford et al., 1958, Baldwin, 1960, Penfield and Perot, 1963, Horowitz et al., 1968, Ferguson et al., 1969, Halgren et al., 1978, Wieser, 1979, Gloor et al., 1982, Fish et al., 1993, Bancaud et al., 1994, Kahane et al., 2003, Barbeau et al., 2005, Vignal et al., 2007, Mulak et al., 2008, Jacobs et al., 2012). Equally important, however, is the specificity of these results: a comprehensive review of cognitive-affective findings from electrophysiological studies in humans, summarizing the results of over 100 such studies conducted over the past eighty years, found that such phenomena 	   150	  were produced almost exclusively by stimulation of medial temporal lobe structures (Selimbeyoglu & Parvizi, 2010; see our summary of relevant results in our Table 25). Stimulation of nearby temporopolar cortex (Penfield and Perot, 1963, Halgren et al., 1978, Bancaud et al., 1994) or lateral temporal cortex (Penfield, 1958, Mullan and Penfield, 1959, Penfield and Perot, 1963, Bancaud et al., 1994) also occasionally elicits such phenomena, but these reports are comparatively rare, and moreover electrical stimulation is known to spread to adjacent cortical areas, making it difficult to strictly rule out a medial temporal origin (or co-activation) in many of these cases (Gloor et al., 1982, Gloor, 1990, Bancaud et al., 1994). Indeed, many of the stimulations to lateral temporal cortex that elicited thought- and dream-like experiences were reported from electrodes at a depth of several centimeters, supporting such a possibility (Selimbeyoglu and Parvizi, 2010). Stimulation of lateral prefrontal cortex (Blanke et al., 2000a) and orbitofrontal cortex (Mahl et al., 1964) can likewise occasionally elicit such experiences, but such reports are few and far between (Table 25). Most striking is that no such experiences appear to have ever been reported from stimulation of virtually any other area in the brain (Selimbeyoglu and Parvizi, 2010), even from the many other regions consistently recruited by self-generated thought in functional neuroimaging investigations. For instance, fewer than 1% of stimulations to the inferior parietal lobule elicit such phenomena (Selimbeyoglu and Parvizi, 2010). We summarize these results in Table 25 and Fig. 22. Although absence of evidence is not necessarily evidence of absence, the results are intriguing given the number of investigations already carried out and the span of time over which such investigations have been taking place (well over a hundred studies, over nearly a hundred years). Together these results strongly suggest that the medial temporal lobe is a key generation 	   151	  site for many forms of self-generated thought, including dreaming – a hypothesis that could be examined more forcefully with lesion patients (see previous section). A central limitation of drawing conclusions based on this body of research is that the results were primarily evoked by electrical stimulation – they mostly do not represent self-generated, spontaneous brain activity giving rise to immersive thoughts and hallucinatory, dream-like experiences. Several studies, however, have passively recorded spontaneous brain activity in conjunction with first-person reports of accompanying experience, and yielded consistent findings: the medial temporal lobe appears to be by far the most common origin of self-generated brain activity accompanied by spontaneous memories, thoughts, and dream-like experiences (Bancaud et al., 1994, Vignal et al., 2007, Gelbard-Sagiv et al., 2008).  	   	  	   152	  	  Figure	  22.	  Preferential involvement of medial temporal lobe structures and temporopolar cortex in electrophysiological stimulations (or spontaneous discharges) eliciting memories, thoughts, or hallucinatory, dream-like experiences. Percentage of stimulations or spontaneous discharges that elicited a first-person experience of memories, thoughts, or hallucinatory, dream-like experiences, based on more than 100 independent investigations. Not shown are data for hundreds of other stimulations throughout the brain, for which no such thought- or dream-like experiences have ever been reported. Only brain areas with ≥10 stimulations or discharges reported in the literature are visualized. Drawn from data in our Table 25, based on data in Supplementary Table 1 of the comprehensive review of Selimbeyoglu & Parvizi (2010). HPC: hippocampus; IPL: inferior parietal lobule; ITG: inferior temporal gyrus; MTG: middle temporal gyrus; PHC: parahippocampal cortex; STG: superior temporal gyrus; TOJ: temporo-occipital junction; TPC: temporopolar cortex. 	  	   	  	   153	  Table	  25.	  Summary	  of	  human	  electrophysiology	  studies	  demonstrating	  elicitation	  of	  memories,	  thoughts,	  or	  hallucinatory,	  dream-­‐like	  experiences.	  Brain Region Stimulations/ discharges elicitinga  Total stimulations/ discharges Percentage eliciting  References Temporal Lobe Hippocampus 25 46 54% (Halgren et al., 1978, Fish et al., 1993, Bancaud et al., 1994, Kahane et al., 2003, Vignal et al., 2007, Mulak et al., 2008) Amygdala 13 36 36% (Ferguson et al., 1969, Halgren et al., 1978, Fish et al., 1993, Vignal et al., 2007) Parahippocampal region 9 16 56% (Feindel and Penfield, 1954, Penfield and Perot, 1963, Vignal et al., 2007) Temporopolar cortex 5 11 45% (Penfield and Perot, 1963, Halgren et al., 1978, Bancaud et al., 1994, Ostrowsky et al., 2002, Mulak et al., 2008) Inferior temporal gyrus 1 21 5% (Penfield and Perot, 1963) Middle temporal gyrus 7 42 17% (Penfield, 1958, Mullan and Penfield, 1959, Penfield and Perot, 1963, Kahane et al., 2003) Superior temporal gyrus 24 99 24% (Mullan and Penfield, 1959, Penfield and Perot, 1963, Morris et al., 1984) Temporo-occipital junction 4 17 24% (Penfield and Perot, 1963, Morris et al., 1984, Lee et al., 2000) Frontal Lobe Inferior frontal gyrus 1 7 14% (Blanke et al., 2000a) Middle frontal gyrus 2 8 25% (Blanke et al., 2000a) Orbitofrontal cortex 1 4 25% (Mahl et al., 1964) Supplementary motor area 1 6 17% (Beauvais et al., 2005) Parietal Lobe Inferior parietal lobule 2 42 .05% (Blanke et al., 2000b, Schulz et al., 2007) 	  Based on data in Supplementary Table 1 in the comprehensive review conducted by Selimbeyoglu & Parvizi (2010). Data for brain areas with ≥10 stimulations/discharges reported in the literature are visualized in Fig. 22.  	   154	  The	  subsequent	  ontogeny	  of	  self-­‐generated	  thought	  Even highly localized stimulation or spontaneous neural firing will tend to spread to proximate and distant brain areas through short- and long-range connections (Halgren and Chauvel, 1992, Selimbeyoglu and Parvizi, 2010). Assuming that the medial temporal lobe is one of the main sources of spontaneous activity giving rise to thoughts and spontaneously recalled memories, the question remains of how and on what timescale this activity spreads, and where in the brain it is most likely to spread to. Given that the medial temporal lobe is well connected to virtually every other part of the brain (Buzsáki, 1989, Rolls, 2000, Simons and Spiers, 2003, Buzsaki, 2006), the possibilities are legion.  Our view of the role of the medial temporal lobe in spontaneously reactivating memory traces, and in recombining mnemic material into novel thoughts and imaginings, draws heavily on the hippocampal indexing theory (Teyler and DiScenna, 1985, 1986, Moscovitch, 1992, Teyler and Rudy, 2007). In essence, “indexing theory proposes that the content of our experiences is stored in the multiple neocortical loci activated by experience and the hippocampus stores an index of those neocortical loci” (Teyler & Rudy, 2007; p. 1160). Indexing theory is primarily concerned with accounting for the reactivation of memories; our view expands on this to account also for novel patterns of thought that draw on mnemic material but recombine it into new thoughts and simulations of experience. If a specific pattern of medial temporal lobe activity and connectivity to neocortex indeed gives rise to memory recall, it follows that slightly altered or novel patterns of hippocampal activity will ‘index’ novel patterns of distributed neocortical activity never before experienced – that is, novel thoughts and simulated experiences. The central problem is to understand “the manner in which electrical stimulation of a particular location [or spontaneous, self-generated activity at a given location] 	   155	  leads to activation of [a] very widespread but, at the same time, very particular network” (Bancaud et al., 1994; p. 87). We therefore do not predict a single, universal pattern of spreading activity. The pattern of subsequent recruitment should instead be related to the experiential qualities of the accompanying thoughts, including the sensory modalities instantiated (visual, auditory, somatosensory, etc.); affective tone (positive, negative, or neutral); and other qualities such as temporal orientation (past, present, future) and goal-relatedness. For instance, thoughts that are visual in nature may involve spreading activity (at the neuronal level) and subsequently observable recruitment (e.g., with functional neuroimaging) of areas such as the lingual gyrus. Similarly, we would predict that more goal-related thoughts involving planning for the future should tend to recruit prefrontal executive areas such as dorsal anterior cingulate cortex and rostrolateral prefrontal cortex. A further consideration is that new thoughts and imagined mental content may not be determined merely by patterns of synaptic connections alone – these anatomical connections could also interact with a host of other factors that all participate in the sculpting of spontaneous and ‘noisy’ brain activity into an overall pattern of activity ultimately corresponding to subjectively experienced mental content of one form or another. Aside from the basic neurochemical state of the brain at a given time (see next section), “current sensory input, cognitive context and/or psychosocial concerns could sometimes be major influences in defining the final pattern to emerge from the sculpting” (Bancaud et al., 1994; p. 87). Evidence bearing on this question is unfortunately much more sparse – commensurate with the increased difficulty of the problem. A comprehensive understanding of the spread of spontaneous activations would require widespread electrode placement throughout the brain – a 	   156	  situation uncommon in human electrophysiology studies, where electrode placement is determined strictly by clinical criteria and rarely requires simultaneous recording from many widely dispersed sites.  To truly explore the hypothesized model of self-generated thought ontogeny requires recording from relevant cortical areas and the medial temporal lobe, simultaneously. The study that probably best approximates this ideal methodology investigated episodic memory retrieval while recording simultaneously from medial temporal lobe and the retrosplenial cortex in the posteromedial area (Foster et al., 2013). The authors reported that episodic memory retrieval involved phase locking in the theta band (3-4 Hz) between these two regions, and that this concerted activity was unique to this frequency band as well as to these two regions. Most intriguing is that the coupling was strongest prior to actual peak high-frequency activity in the retrosplenial cortex, potentially suggesting a primary role for the medial temporal lobe. Another study investigating spontaneous memory recall with human intracranial electrophysiology found that medial temporal lobe structures demonstrated some of the strongest increases in high-frequency (i.e., high 𝛾–band) activity just prior to conscious recall and reporting of the memory (Burke et al., 2014), whereas high-frequency activity peaked later in several parietal, temporal, and frontal regions (see their Fig. 4). Importantly, although the medial temporal lobe was not the only region in which high-frequency activity peaked prior to recall, it was the only one in which this activity significantly predicted subsequent memory recall, suggesting a cardinal role in memory recollection (Burke et al., 2014). A limitation of these studies (for our purposes) is that they involve memory retrieval in one form or another; therefore they do not speak to the possibility of truly novel patterns of activity – a precondition for any neuron-level model of self-generated thought. Some preliminary 	   157	  evidence for this possibility, however, comes from animal models. Several well-known animal studies have demonstrated that patterns of medial temporal lobe activity reflecting recent spatiotemporal experiences and memories are spontaneously replayed during periods of resting wakefulness (Foster and Wilson, 2006, Diba and Buzsáki, 2007) or subsequent sleep (Wilson and McNaughton, 1994). Coordinated replay across medial temporal lobe and various neocortical areas, including visual regions (Ji and Wilson, 2007) and posterior parietal cortex (Qin et al., 1997), has also been reported – although a causal role for the medial temporal lobe in initiating this activity has not been shown. Interestingly, recent studies have shown that ‘replay’ need not be a mere recapitulation of previous firing patterns, but often contains novel firing sequences that do not correspond to any actual spatiotemporal sequence of experience (Davidson et al., 2009, Gupta et al., 2010). These novel firing sequences have been interpreted as evidence for planning and imagining of alternative behaviors (Knierim, 2009, Gupta et al., 2010) – i.e., the generation of novel thoughts, plans, and imagined scenarios, akin to self-generated thought content in humans. In summary, although the network-level ontogeny of thoughts putatively generated in the medial temporal lobe remains largely obscure, many important prerequisites for an ‘indexing’ scenario have already been demonstrated. The medial temporal lobe is known to spontaneously reactivate patterns of activity first instantiated during novel experience and learning, and this reactivation takes place not only during sleep but also waking behavior and restful states. This patterned replay can be temporally synchronized with various neocortical brain areas (Qin et al., 1997, Ji and Wilson, 2007), and, critically, ‘replay’ can in fact involve novel patterns of activity that do not correspond to any specific experience (Davidson et al., 2009, Knierim, 2009, Gupta et al., 2010). Although still far from definitive, all of these findings are consistent with an 	   158	  ‘indexing’ theory of self-generated thought origin and ontogeny, whereby spontaneous medial temporal lobe activity activates neuronal networks dispersed throughout the brain, instantiating either recall of memories or the experience of novel thoughts and imaginings. Perhaps most important, electrode montages that include both medial temporal lobe and other regions clearly important to self-generated thought are occasionally employed in clinical settings with human participants (Ekstrom et al., 2003, Gelbard-Sagiv et al., 2008, Foster et al., 2013, Burke et al., 2014); the stage is therefore set for further exploration of the temporal and spatial dynamics that characterize the ontogeny of thoughts in the human brain. The most fruitful approach might be to combine intracranial electrophysiology (which has unparalleled temporal resolution, but cannot be expected to cover a large number of brain regions in human studies) with functional MRI (with poor temporal resolution, but the ability to sample from all brain regions simultaneously). The safe and optimal combination of these methods is an area of active research and has already been demonstrated in feasibility studies (Carmichael et al., 2007, Carmichael et al., 2010).	  	   	  	    	   159	  CHAPTER	  9	  –	  SUMMARY	  AND	  CONCLUSIONS	    In this final chapter we briefly summarize the key findings from each of the seven preceding chapters and attempt to integrate them to draw some general conclusions about the neural basis of self-generated thought.   Self-­‐generated	  thought	  in	  general	  is	  associated	  with	  widespread	  brain	  recruitment	   Two central conclusions can be drawn from our meta-analytic assessment of general brain recruitment associated with various forms of self-generated thought (Experiment 1). First, as was long suspected, the default network participates critically in self-generated thought: every major default network region shows consistent recruitment. However, this is not the whole story: numerous other brain areas and networks also show consistent recruitment, most notably frontoparietal control network areas, including dorsal anterior cingulate cortex and lateral prefrontal regions. Other areas beyond both these networks are also activated consistently, including medial occipital areas, temporopolar cortex, and the insula. This wide variety of brain recruitment accords well with the subjective diversity of self-generated thought, but cries out for more detailed study to resolve the specific role played by these various regions and networks – a major aim of our subsequent Experiments 3, 5, and 6.   	  	   	  	   160	  Self-­‐generated	  thought	  and	  imagery	  in	  sleep	  and	  dreaming	  share	  a	  neural	  substrate	  with	  waking	  thought	  Despite the apparent quiescence of sleep, it is now established that all stages of sleep are accompanied to some degree by self-generated thought and imagery (Nielsen, 2000). The relative frequency of this sleep mentation differs markedly between stages, peaking in REM sleep and bottoming out in SWS. Our meta-analyses of all functional neuroimaging investigations that have explored the various stages of sleep revealed intriguing correspondences between the relative frequency of such self-generated thought and imagery, and activation in default and visual network areas, as expected based on our earlier work along these lines (Fox et al., 2013, Domhoff and Fox, 2015). We also found a similar correspondence between deactivation of frontoparietal control networks regions and the corresponding loss of cognitive control and metacognitive awareness in most sleep stages (with the notable exception of lucid REM sleep, during which activation in these regions is restored). Aside from arguing forcefully that self-generated thought shares a common neural basis regardless of conscious state (sleeping or awake), these results show that self-generated thought can easily take place without executive system recruitment, therefore setting the stage for our subsequent experiments that aimed to specifically investigate the role played by executive regions (Experiments 3 and 5).   Distinct	  categories	  of	  thought	  can	  be	  resolved	  at	  the	  neural	  level	  with	  fMRI	  Our fMRI results point to the ability of functional neuroimaging to resolve specific categories of thought at the neural level, based merely on large numbers of first-person reports. Although we concluded that, despite our large number of thought probes (n = 120), our analyses were still under-powered, nonetheless important distinctions were drawn between basic neural 	   161	  correlates for several major categories of thought. The most notable findings, we believe, are the first direct evidence for brain recruitment during self-generated thought involving executive processes, namely intentionally directed/guided thought, and thought related to one’s goals and current concerns. Functional neuroimaging revealed executive system recruitment directly associated with these types of thought (as well as other brain recruitment) – providing the first tentative answer to questions about what role executive regions might be playing in apparently spontaneous and undirected mental processes at rest. These results are further bolstered by our subsequent morphometric neuroimaging analyses, which found concordant results.  The	  MRI	  scanner	  environment	  does	  not	  appreciably	  affect	  self-­‐generated	  thought	  content	  The central conclusion of Experiment 4 (Chapter 5) is that the MRI scanner environment does not appear to have any appreciable influence on the overall content or valence of self-generated thought. The rate of stimulus-independent thought might be somewhat higher in the MRI scanner compared to a laboratory setting, but this observation might also have been due to chance differences in stimulus-independent thought frequency between our two samples.  These findings apply at the group level (i.e., average thought content does not differ systematically across environmental contexts). A further conclusion, however, can be drawn at the individual level, based on data from our participants who completed the thought sampling paradigm both in and outside of the MRI scanner. These participants, too, showed overall highly stable patterns of thought both in and outside the MRI scanner, and across several days. These findings broadly suggest that a ~1 hr thought sampling session involving >100 thought probes can provide reasonably stable, trait-like measures of individual thought propensity that could 	   162	  conceivably be used as individual difference measures in further analyses – a task we took up in our subsequent Experiments 5 and 6. 	  	  	  Interindividual	  differences	  in	  thought	  content	  are	  associated	  with	  individual	  neuroanatomical	  heterogeneity	  	   Converting our large number of thought probes into individual difference measures, we examined the possibility of relationships between individual neuroanatomical heterogeneities and subject-specific patterns of thought content. We found many such relationships using measures of both grey matter concentration as well as white matter fractional anisotropy. The most intriguing results from these analyses again pointed to the specific role played by executive brain regions in intentionally directing or guiding the flow of self-generated thoughts, and independently, in facilitating thought related to one’s goals and current concerns. Given our relatively small samples in these correlational analyses, however, and our ongoing collection of participants, these results should be interpreted with some caution for the time being. 	  	  Lesions	  studies	  point	  to	  necessary	  neural	  substrates	  of	  self-­‐generated	  thought	  	  	   A synthesis of nearly a century of neuropsychological lesion studies provided telling evidence that only certain regions are absolutely critical to self-generated thought. Reviewing this literature, we found that self-generated thought is abolished, or severely attenuated, with lesions to only three brain regions: the medial prefrontal cortex, the inferior parietal lobule/temporoparietal junction, and the medial temporal lobe. Additionally, medial occipital lesions abolish the visual imagery so characteristic of self-generated thought in both waking and 	   163	  sleeping (see Chapters 1 and 3). The selectivity of these findings is bolstered by the complementary discovery that lesions to many other brain regions have no appreciable effect on self-generated thought frequency or content. 	  	  Human	  intracranial	  electrophysiology	  suggests	  a	  medial	  temporal	  lobe	  origin	  site	  for	  self-­‐generated	  thought	  	   The preceding work addressed many functional neural qualities of self-generated thought, but none of it addressed perhaps the most intriguing question of all: where in the brain our self-generated mental content originates. Synthesizing nearly one hundred years of isolated human intracranial electrophysiology reports, we concluded that there is strong evidence that the medial temporal lobe is the major (if not necessarily the only) origin site of most self-generated thought. Very little research, however, has addressed the ways in which this self-generated medial temporal lobe activity then propagates to, and recruits, other brain regions – a major avenue for future research. 	   	  	   164	  REFERENCES	  	  	  Addis	  DR,	  Pan	  L,	  Vu	  MA,	  Laiser	  N,	  Schacter	  DL	  (2009)	  Constructive	  episodic	  simulation	  of	  the	  future	  and	  the	  past:	  distinct	  subsystems	  of	  a	  core	  brain	  network	  mediate	  imagining	  and	  remembering.	  Neuropsychologia	  47:2222-­‐2238.	  Allen	  M,	  Smallwood	  J,	  Christensen	  J,	  Gramm	  D,	  Rasmussen	  B,	  Gaden-­‐Jensen	  C,	  Roepstorff	  A,	  Lutz	  A	  (2013)	  The	  balanced	  mind:	  the	  variability	  of	  task-­‐unrelated	  thoughts	  predicts	  error-­‐monitoring.	  Frontiers	  in	  Human	  Neuroscience	  7.	  Altman	  J,	  Das	  GD	  (1965)	  Autoradiographic	  and	  histological	  evidence	  of	  postnatal	  hippocampal	  neurogenesis	  in	  rats.	  Journal	  of	  Comparative	  Neurology	  124:319-­‐335.	  Anālayo	  (2003)	  Satipaṭṭhāna:	  The	  Direct	  Path	  to	  Realization:	  Windhorse	  Publications.	  Andelman	  F,	  Hoofien	  D,	  Goldberg	  I,	  Aizenstein	  O,	  Neufeld	  MY	  (2010)	  Bilateral	  hippocampal	  lesion	  and	  a	  selective	  impairment	  of	  the	  ability	  for	  mental	  time	  travel.	  Neurocase	  16:426-­‐435.	  Anderson	  BJ	  (2011)	  Plasticity	  of	  gray	  matter	  volume:	  the	  cellular	  and	  synaptic	  plasticity	  that	  underlies	  volumetric	  change.	  Developmental	  psychobiology	  53:456-­‐465.	  Andersson	  JL,	  Jenkinson	  M,	  Smith	  S	  (2007a)	  Non-­‐linear	  optimisation.	  FMRIB	  technical	  report	  TR07JA1.	  University	  of	  Oxford	  FMRIB	  Centre:	  Oxford,	  UK.	  Andersson	  JL,	  Jenkinson	  M,	  Smith	  S	  (2007b)	  Non-­‐linear	  registration,	  aka	  Spatial	  normalisation	  FMRIB	  technical	  report	  TR07JA2.	  FMRIB	  Analysis	  Group	  of	  the	  University	  of	  Oxford.	  	   165	  Andreasen	  NC,	  O'Leary	  DS,	  Cizaldo	  T,	  Arndt	  S,	  Rezai	  K,	  al.	  e	  (1995)	  Remembering	  the	  past:	  two	  facets	  of	  episodic	  memory	  explored	  with	  positron	  emission	  tomography.	  American	  Journal	  of	  Psychiatry	  152.	  Andrews-­‐Hanna	  JR	  (2012)	  The	  brain's	  default	  network	  and	  its	  adaptive	  role	  in	  internal	  mentation.	  Neuroscientist	  18:251-­‐270.	  Andrews-­‐Hanna	  JR,	  Kaiser	  RH,	  Turner	  AEJ,	  Reineberg	  A,	  Godinez	  D,	  Dimidjian	  S,	  Banich	  M	  (2013)	  A	  penny	  for	  your	  thoughts:	  Dimensions	  of	  thought	  content	  and	  relationships	  with	  individual	  differences	  in	  emotional	  well-­‐being.	  Frontiers	  in	  Perception	  Science.	  Andrews-­‐Hanna	  JR,	  Reidler	  JS,	  Huang	  C,	  Buckner	  RL	  (2010)	  Evidence	  for	  the	  default	  network's	  role	  in	  spontaneous	  cognition.	  Journal	  of	  neurophysiology	  104:322-­‐335.	  Andrews-­‐Hanna	  JR,	  Smallwood	  J,	  Spreng	  RN	  (2014)	  The	  default	  network	  and	  self-­‐generated	  thought:	  Component	  processes	  and	  dynamic	  control.	  Annals	  of	  the	  New	  York	  Academy	  of	  Sciences	  1316:29-­‐52.	  Anticevic	  A,	  Cole	  MW,	  Murray	  JD,	  Corlett	  PR,	  Wang	  XJ,	  Krystal	  JH	  (2012)	  The	  role	  of	  default	  network	  deactivation	  in	  cognition	  and	  disease.	  Trends	  Cogn	  Sci	  16:584-­‐592.	  Antonova	  E,	  Sharma	  T,	  Morris	  R,	  Kumari	  V	  (2004)	  The	  relationship	  between	  brain	  structure	  and	  neurocognition	  in	  schizophrenia:	  a	  selective	  review.	  Schizophrenia	  research	  70:117-­‐145.	  Antrobus	  J	  (1991)	  Dreaming:	  Cognitive	  processes	  during	  cortical	  activation	  and	  high	  afferent	  thresholds.	  Psychological	  Review	  98:96.	  Aristoteles,	  Gallop	  D	  (1996)	  Aristotle	  on	  Sleep	  and	  Dreams:	  A	  Text	  and	  Translation:	  Cambridge	  Univ	  Press.	  	   166	  Aserinsky	  E,	  Kleitman	  N	  (1953)	  Regularly	  occurring	  periods	  of	  eye	  motility,	  and	  concomitant	  phenomena,	  during	  sleep.	  Science	  118:273-­‐274.	  Aserinsky	  E,	  Kleitman	  N	  (1955)	  A	  motility	  cycle	  in	  sleeping	  infants	  as	  manifested	  by	  ocular	  and	  gross	  bodily	  activity.	  Journal	  of	  Applied	  Physiology	  8:11-­‐18.	  Ashburner	  J	  (2007)	  A	  fast	  diffeomorphic	  image	  registration	  algorithm.	  NeuroImage	  38:95-­‐113.	  Ashburner	  J,	  Friston	  KJ	  (2000)	  Voxel-­‐based	  morphometry—the	  methods.	  NeuroImage	  11:805-­‐821.	  Baird	  B,	  Smallwood	  J,	  Schooler	  JW	  (2011)	  Back	  to	  the	  future:	  Autobiographical	  planning	  and	  the	  functionality	  of	  mind-­‐wandering.	  Consciousness	  and	  cognition	  20:1604-­‐1611.	  Baldwin	  M	  (1960)	  Electrical	  stimulation	  of	  the	  mesial	  temporal	  region.	  Electrical	  studies	  on	  the	  unanesthetized	  brain	  159-­‐176.	  Bancaud	  J,	  Brunet-­‐Bourgin	  F,	  Chauvel	  P,	  Halgren	  E	  (1994)	  Anatomical	  origin	  of	  deja	  vu	  and	  vivid'memories'	  in	  human	  temporal	  lobe	  epilepsy.	  Brain	  117:71-­‐90.	  Barbeau	  E,	  Wendling	  F,	  Régis	  J,	  Duncan	  R,	  Poncet	  M,	  Chauvel	  P,	  Bartolomei	  F	  (2005)	  Recollection	  of	  vivid	  memories	  after	  perirhinal	  region	  stimulations:	  synchronization	  in	  the	  theta	  range	  of	  spatially	  distributed	  brain	  areas.	  Neuropsychologia	  43:1329-­‐1337.	  Barbera	  J	  (2008)	  Sleep	  and	  dreaming	  in	  Greek	  and	  Roman	  philosophy.	  Sleep	  medicine	  9:906-­‐910.	  Barrett	  D	  (1991)	  Flying	  dreams	  and	  lucidity:	  An	  empirical	  study	  of	  their	  relationship.	  Dreaming	  1:129.	  	   167	  Beauvais	  K,	  Biraben	  A,	  Seigneuret	  E,	  Saïkali	  S,	  Scarabin	  J-­‐M	  (2005)	  Subjective	  signs	  in	  premotor	  epilepsy:	  confirmation	  by	  stereo-­‐electroencephalography.	  Epileptic	  disorders	  7:347-­‐354.	  Bellebaum	  C,	  Daum	  I	  (2007)	  Cerebellar	  involvement	  in	  executive	  control.	  The	  Cerebellum	  6:184-­‐192.	  Berman	  MG,	  Peltier	  S,	  Nee	  DE,	  Kross	  E,	  Deldin	  PJ,	  Jonides	  J	  (2010)	  Depression,	  rumination	  and	  the	  default	  network.	  Social	  cognitive	  and	  affective	  neuroscience	  nsq080.	  Bernhardt	  BC,	  Smallwood	  J,	  Tusche	  A,	  Ruby	  FJ,	  Engen	  HG,	  Steinbeis	  N,	  Singer	  T	  (2014)	  Medial	  prefrontal	  and	  anterior	  cingulate	  cortical	  thickness	  predicts	  shared	  individual	  differences	  in	  self-­‐generated	  thought	  and	  temporal	  discounting.	  NeuroImage	  90:290-­‐297.	  Berntsen	  D,	  Rubin	  DC	  (2008)	  The	  reappearance	  hypothesis	  revisited:	  Recurrent	  involuntary	  memories	  after	  traumatic	  events	  and	  in	  everyday	  life.	  Memory	  &	  Cognition	  36:449-­‐460.	  Bickford	  RG,	  Mulder	  DW,	  Dodge	  Jr	  HW,	  Svien	  HJ,	  Rome	  HP	  (1958)	  Changes	  in	  memory	  function	  produced	  by	  electrical	  stimulation	  of	  the	  temporal	  lobe	  in	  man.	  Research	  Publications	  of	  the	  Association	  for	  Research	  in	  Nervous	  &	  Mental	  Disease.	  Binder	  JR,	  Frost	  JA,	  Hammeke	  TA,	  Bellgowan	  PSF,	  Rao	  SM,	  Cox	  RW	  (1999)	  Conceptual	  processing	  during	  the	  conscious	  resting	  state:	  A	  functional	  MRI	  study.	  Journal	  of	  Cognitive	  Neuroscience	  11:80-­‐93.	  Bischof	  M,	  Bassetti	  CL	  (2004)	  Total	  dream	  loss:	  a	  distinct	  neuropsychological	  dysfunction	  after	  bilateral	  PCA	  stroke.	  Annals	  of	  Neurology	  56:583-­‐586.	  	   168	  Blanke	  O,	  Landis	  T,	  Seeck	  M	  (2000a)	  Electrical	  cortical	  stimulation	  of	  the	  human	  prefrontal	  cortex	  evokes	  complex	  visual	  hallucinations.	  Epilepsy	  &	  Behavior	  1:356-­‐361.	  Blanke	  O,	  Perrig	  S,	  Thut	  G,	  Landis	  T,	  Seeck	  M	  (2000b)	  Simple	  and	  complex	  vestibular	  responses	  induced	  by	  electrical	  cortical	  stimulation	  of	  the	  parietal	  cortex	  in	  humans.	  Journal	  of	  Neurology,	  Neurosurgery	  &	  Psychiatry	  69:553-­‐556.	  Borbély	  AA,	  Baumann	  F,	  Brandeis	  D,	  Strauch	  I,	  Lehmann	  D	  (1981)	  Sleep	  deprivation:	  effect	  on	  sleep	  stages	  and	  EEG	  power	  density	  in	  man.	  Electroencephalography	  and	  clinical	  neurophysiology	  51:483-­‐493.	  Braun	  A,	  Balkin	  T,	  Wesensten	  N,	  Carson	  R,	  Varga	  M,	  Baldwin	  P,	  Selbie	  S,	  Belenky	  G,	  Herscovitch	  P	  (1997)	  Regional	  cerebral	  blood	  flow	  throughout	  the	  sleep-­‐wake	  cycle.	  Brain	  120:1173-­‐1197.	  Brewer	  JA,	  Worhunsky	  PD,	  Gray	  JR,	  Tang	  Y,	  Weber	  J,	  Kober	  H	  (2011)	  Meditation	  experience	  is	  associated	  with	  differences	  in	  default	  mode	  network	  activity	  and	  connectivity.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  USA	  108:20254-­‐20259.	  Broca	  PP	  (1861)	  Perte	  de	  la	  parole,	  ramollissement	  chronique	  et	  destruction	  partielle	  du	  lobe	  antérieur	  gauche	  du	  cerveau.	  Bulletin	  de	  la	  Société	  Anthropologique	  2:235-­‐238.	  Buchsbaum	  MS,	  Gillin	  JC,	  Wu	  J,	  Hazlett	  E,	  Sicotte	  N,	  Dupont	  RM,	  Bunney	  WE	  (1989)	  Regional	  cerebral	  glucose	  metabolic	  rate	  in	  human	  sleep	  assessed	  by	  positron	  emission	  tomography.	  Life	  sciences	  45:1349-­‐1356.	  Buckner	  RL,	  Andrews-­‐Hanna	  JR,	  Schacter	  DL	  (2008a)	  The	  Brain's	  Default	  Network:	  Anatomy,	  Function,	  and	  Relevance	  to	  Disease.	  Annals	  of	  the	  New	  York	  Academy	  of	  Sciences	  1124:1-­‐38.	  	   169	  Buckner	  RL,	  Andrews-­‐Hanna	  JR,	  Schacter	  DL	  (2008b)	  The	  brain's	  default	  network:	  anatomy,	  function,	  and	  relevance	  to	  disease.	  Ann	  N	  Y	  Acad	  Sci	  1124:1-­‐38.	  Bunge	  M	  (1998)	  Philosophy	  of	  science:	  from	  problem	  to	  theory:	  Transaction	  Publishers.	  Burke	  JF,	  Sharan	  AD,	  Sperling	  MR,	  Ramayya	  AG,	  Evans	  JJ,	  Healey	  MK,	  Beck	  EN,	  Davis	  KA,	  Lucas	  TH,	  Kahana	  MJ	  (2014)	  Theta	  and	  high-­‐frequency	  activity	  mark	  spontaneous	  recall	  of	  episodic	  memories.	  The	  Journal	  of	  Neuroscience	  34:11355-­‐11365.	  Buzsaki	  G	  (2006)	  Rhythms	  of	  the	  Brain:	  Oxford	  University	  Press.	  Buzsáki	  G	  (1989)	  Two-­‐stage	  model	  of	  memory	  trace	  formation:	  a	  role	  for	  “noisy”	  brain	  states.	  Neuroscience	  31:551-­‐570.	  Buzsáki	  G,	  Anastassiou	  CA,	  Koch	  C	  (2012)	  The	  origin	  of	  extracellular	  fields	  and	  currents—EEG,	  ECoG,	  LFP	  and	  spikes.	  Nature	  Reviews	  Neuroscience	  13:407-­‐420.	  Cabeza	  R,	  Ciaramelli	  E,	  Olson	  IR,	  Moscovitch	  M	  (2008)	  The	  parietal	  cortex	  and	  episodic	  memory:	  an	  attentional	  account.	  Nature	  Reviews	  Neuroscience	  9:613-­‐625.	  Cacioppo	  JT,	  Tassinary	  LG	  (1990)	  Inferring	  psychological	  significance	  from	  physiological	  signals.	  American	  Psychologist	  45:16.	  Campbell	  DT	  (1960)	  Blind	  variation	  and	  selective	  retention	  in	  creative	  thought	  as	  in	  other	  knowledge	  processes.	  Psychological	  Review	  67:380-­‐400.	  Capra	  F,	  Luisi	  PL	  (2014)	  The	  systems	  view	  of	  life:	  A	  unifying	  vision:	  Cambridge	  University	  Press.	  Carmichael	  DW,	  Pinto	  S,	  Limousin-­‐Dowsey	  P,	  Thobois	  S,	  Allen	  PJ,	  Lemieux	  L,	  Yousry	  T,	  Thornton	  JS	  (2007)	  Functional	  MRI	  with	  active,	  fully	  implanted,	  deep	  brain	  stimulation	  systems:	  safety	  and	  experimental	  confounds.	  NeuroImage	  37:508-­‐517.	  	   170	  Carmichael	  DW,	  Thornton	  JS,	  Rodionov	  R,	  Thornton	  R,	  McEvoy	  AW,	  Ordidge	  RJ,	  Allen	  PJ,	  Lemieux	  L	  (2010)	  Feasibility	  of	  simultaneous	  intracranial	  EEG-­‐fMRI	  in	  humans:	  a	  safety	  study.	  NeuroImage	  49:379-­‐390.	  Cartwright	  RD,	  Lloyd	  S,	  Knight	  S,	  Trenholme	  I	  (1984)	  Broken	  dreams:	  A	  study	  of	  the	  effects	  of	  divorce	  and	  depression	  on	  dream	  content.	  Psychiatry	  47:251-­‐259.	  Cavallero	  C,	  Cicogna	  P,	  Natale	  V,	  Occhionero	  M	  (1992)	  Slow	  wave	  sleep	  dreaming.	  Sleep:	  Journal	  of	  Sleep	  Research	  &	  Sleep	  Medicine.	  Christoff	  K	  (2012)	  Undirected	  thought:	  neural	  determinants	  and	  correlates.	  Brain	  Res	  1428:51-­‐59.	  Christoff	  K,	  Gordon	  AM,	  Smallwood	  J,	  Smith	  R,	  Schooler	  JW	  (2009)	  Experience	  sampling	  during	  fMRI	  reveals	  default	  network	  and	  executive	  system	  contributions	  to	  mind	  wandering.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  of	  the	  United	  States	  of	  America	  106:8719-­‐8724.	  Christoff	  K,	  Irving	  ZC,	  Fox	  KCR,	  Spreng	  RN,	  Andrews-­‐Hanna	  JR	  (in	  press)	  Mind-­‐wandering	  as	  spontaneous	  thought:	  A	  dynamic	  framework.	  Nature	  Reviews	  Neuroscience.	  Christoff	  K,	  Owen	  AM	  (2006)	  Improving	  reverse	  neuroimaging	  inference:	  Cognitive	  domain	  versus	  cognitive	  complexity.	  Trends	  in	  Cognitive	  Sciences	  10:352-­‐353.	  Christoff	  K,	  Ream	  JM,	  Gabrieli	  JDE	  (2004)	  Cognitive	  and	  neural	  basis	  of	  spontaneous	  thought	  processes.	  Cortex	  40:623-­‐630.	  Cicogna	  P,	  Natale	  V,	  Occhionero	  M,	  Bosinelli	  M	  (1998)	  A	  comparison	  of	  mental	  activity	  during	  sleep	  onset	  and	  morning	  awakening.	  Sleep:	  Journal	  of	  Sleep	  Research	  &	  Sleep	  Medicine.	  	   171	  Clark	  A	  (2013)	  Whatever	  next?	  Predictive	  brains,	  situated	  agents,	  and	  the	  future	  of	  cognitive	  science.	  Behavioral	  and	  Brain	  Sciences	  36:181-­‐204.	  Cohen	  J	  (1992)	  A	  power	  primer.	  Psychological	  Bulletin	  112:155.	  Craig	  A	  (2009)	  How	  do	  you	  feel—now?	  the	  anterior	  insula	  and	  human	  awareness.	  Nature	  Reviews	  Neuroscience	  10.	  Cumming	  G	  (2012)	  Understanding	  the	  new	  statistics:	  Effect	  sizes,	  confidence	  intervals,	  and	  meta-­‐analysis:	  Routledge.	  Cumming	  G	  (2013)	  The	  new	  statistics	  why	  and	  how.	  Psychological	  Science	  0956797613504966.	  D'Argembeau	  A,	  Collette	  F,	  Van	  der	  Linden	  M,	  Laureys	  S,	  Del	  Fiore	  G,	  Degueldre	  C,	  Luxen	  A,	  Salmon	  E	  (2005)	  Self-­‐referential	  reflective	  activity	  and	  its	  relationship	  with	  rest:	  a	  PET	  study.	  NeuroImage	  25:616-­‐624.	  Davidson	  TJ,	  Kloosterman	  F,	  Wilson	  MA	  (2009)	  Hippocampal	  replay	  of	  extended	  experience.	  Neuron	  63:497-­‐507.	  Delamillieure	  P,	  Doucet	  G,	  Mazoyer	  B,	  Turbelin	  M-­‐R,	  Delcroix	  N,	  Mellet	  E,	  Zago	  L,	  Crivello	  F,	  Petit	  L,	  Tzourio-­‐Mazoyer	  N	  (2010)	  The	  resting	  state	  questionnaire:	  an	  introspective	  questionnaire	  for	  evaluation	  of	  inner	  experience	  during	  the	  conscious	  resting	  state.	  Brain	  Research	  Bulletin	  81:565-­‐573.	  Dement	  W,	  Kleitman	  N	  (1957a)	  Cyclic	  variations	  in	  EEG	  during	  sleep	  and	  their	  relation	  to	  eye	  movements,	  body	  motility,	  and	  dreaming.	  Electroencephalography	  and	  clinical	  neurophysiology	  9:673-­‐690.	  	   172	  Dement	  W,	  Kleitman	  N	  (1957b)	  The	  relation	  of	  eye	  movements	  during	  sleep	  to	  dream	  activity:	  an	  objective	  method	  for	  the	  study	  of	  dreaming.	  Journal	  of	  experimental	  psychology	  53:339.	  Desseilles	  M,	  Dang-­‐Vu	  TT,	  Sterpenich	  V,	  Schwartz	  S	  (2011)	  Cognitive	  and	  emotional	  processes	  during	  dreaming:	  a	  neuroimaging	  view.	  Consciousness	  and	  cognition	  20:998-­‐1008.	  Deutsch	  E,	  Dalvi	  R	  (2004)	  The	  essential	  Vedanta:	  a	  new	  source	  book	  of	  Advaita	  Vedanta:	  World	  Wisdom	  Books.	  DeYoung	  CG,	  Hirsh	  JB,	  Shane	  MS,	  Papademetris	  X,	  Rajeevan	  N,	  Gray	  JR	  (2010)	  Testing	  predictions	  from	  personality	  neuroscience	  brain	  structure	  and	  the	  big	  five.	  Psychological	  Science.	  Diaz	  BA,	  Van	  Der	  Sluis	  S,	  Moens	  S,	  Benjamins	  JS,	  Migliorati	  F,	  Stoffers	  D,	  Den	  Braber	  A,	  Poil	  S-­‐S,	  Hardstone	  R,	  Van't	  Ent	  D	  (2013)	  The	  Amsterdam	  Resting-­‐State	  Questionnaire	  reveals	  multiple	  phenotypes	  of	  resting-­‐state	  cognition.	  Frontiers	  in	  Human	  Neuroscience	  7.	  Diba	  K,	  Buzsáki	  G	  (2007)	  Forward	  and	  reverse	  hippocampal	  place-­‐cell	  sequences	  during	  ripples.	  Nature	  Neuroscience	  10:1241-­‐1242.	  Dixon	  ML,	  Fox	  KCR,	  Christoff	  K	  (2014)	  A	  framework	  for	  understanding	  the	  relationship	  between	  externally	  and	  internally	  directed	  cognition.	  Neuropsychologia	  62:321-­‐330.	  Domhoff	  GW	  (2003)	  The	  scientific	  study	  of	  dreams:	  Neural	  networks,	  cognitive	  development,	  and	  content	  analysis:	  American	  Psychological	  Association.	  	   173	  Domhoff	  GW	  (2011)	  The	  neural	  substrate	  for	  dreaming:	  is	  it	  a	  subsystem	  of	  the	  default	  network?	  Consciousness	  and	  cognition	  20:1163-­‐1174.	  Domhoff	  GW,	  Fox	  KCR	  (2015)	  Dreaming	  and	  the	  default	  network:	  A	  review,	  synthesis,	  and	  counterintuitive	  research	  proposal.	  Consciousness	  and	  cognition	  33:342-­‐353.	  Dorus	  E,	  Dorus	  W,	  Rechtschaffen	  A	  (1971)	  The	  incidence	  of	  novelty	  in	  dreams.	  Archives	  of	  General	  Psychiatry	  25:364-­‐368.	  Dresler	  M,	  Koch	  SP,	  Wehrle	  R,	  Spoormaker	  VI,	  Holsboer	  F,	  Steiger	  A,	  Sämann	  PG,	  Obrig	  H,	  Czisch	  M	  (2011)	  Dreamed	  movement	  elicits	  activation	  in	  the	  sensorimotor	  cortex.	  Current	  Biology	  21:1833-­‐1837.	  Dresler	  M,	  Wehrle	  R,	  Spoormaker	  VI,	  Koch	  SP,	  Holsboer	  F,	  Steiger	  A,	  Obrig	  H,	  Sämann	  PG,	  Czisch	  M	  (2012)	  Neural	  correlates	  of	  dream	  lucidity	  obtained	  from	  contrasting	  lucid	  versus	  non-­‐lucid	  REM	  sleep:	  a	  combined	  EEG/fMRI	  case	  study.	  Sleep	  35:1017.	  Duff	  MC,	  Kurczek	  J,	  Rubin	  R,	  Cohen	  NJ,	  Tranel	  D	  (2013)	  Hippocampal	  amnesia	  disrupts	  creative	  thinking.	  Hippocampus	  23:1143-­‐1149.	  Dumontheil	  I,	  Gilbert	  SJ,	  Frith	  CD,	  Burgess	  PW	  (2010)	  Recruitment	  of	  lateral	  rostral	  prefrontal	  cortex	  in	  spontaneous	  and	  task-­‐related	  thoughts.	  Q	  J	  Exp	  Psychol	  (Hove)	  63:1740-­‐1756.	  DuPre	  E,	  Spreng	  RN	  (in	  press)	  Depressive	  rumination	  and	  the	  wandering	  mind.	  In:	  The	  Oxford	  Handbook	  of	  Spontaneous	  Thought:	  Mind-­‐wandering,	  Creativity,	  Dreaming,	  and	  Clinical	  Conditions(Fox,	  K.	  C.	  R.	  and	  Christoff,	  K.,	  eds)	  New	  York:	  Oxford	  University	  Press.	  Duvernoy	  HM,	  Cabanis	  EA,	  Bourgouin	  P	  (1991)	  The	  human	  brain:	  surface,	  three-­‐dimensional	  sectional	  anatomy	  and	  MRI:	  Springer	  Verlag	  Wien.	  	   174	  Ehlers	  A,	  Hackmann	  A,	  Michael	  T	  (2004)	  Intrusive	  re-­‐experiencing	  in	  post-­‐traumatic	  stress	  disorder:	  Phenomenology,	  theory,	  and	  therapy.	  Memory	  &	  Cognition	  12:403-­‐415.	  Eickhoff	  SB,	  Bzdok	  D,	  Laird	  AR,	  Kurth	  F,	  Fox	  PT	  (2012)	  Activation	  likelihood	  estimation	  meta-­‐analysis	  revisited.	  NeuroImage	  59:2349-­‐2361.	  Eickhoff	  SB,	  Laird	  AR,	  Grefkes	  C,	  Wang	  LE,	  Zilles	  K,	  Fox	  PT	  (2009)	  Coordinate-­‐based	  activation	  likelihood	  estimation	  meta-­‐analysis	  of	  neuroimaging	  data:	  a	  random-­‐effects	  approach	  based	  on	  empirical	  estimates	  of	  spatial	  uncertainty.	  Hum	  Brain	  Mapp	  30:2907-­‐2926.	  Eifring	  H	  (2015)	  Meditation	  and	  Culture:	  The	  Interplay	  of	  Practice	  and	  Context:	  Bloomsbury	  Publishing.	  Ekstrom	  AD,	  Kahana	  MJ,	  Caplan	  JB,	  Fields	  TA,	  Isham	  EA,	  Newman	  EL,	  Fried	  I	  (2003)	  Cellular	  networks	  underlying	  human	  spatial	  navigation.	  Nature	  425:184-­‐188.	  Ellamil	  M,	  Dobson	  C,	  Beeman	  M,	  Christoff	  K	  (2012)	  Evaluative	  and	  generative	  modes	  of	  thought	  during	  the	  creative	  process.	  NeuroImage	  59:1783-­‐1794.	  Ellamil	  M,	  Fox	  KC,	  Dixon	  ML,	  Pritchard	  S,	  Todd	  RM,	  Thompson	  E,	  Christoff	  K	  (2016)	  Dynamics	  of	  neural	  recruitment	  surrounding	  the	  spontaneous	  arising	  of	  thoughts	  in	  experienced	  mindfulness	  practitioners.	  NeuroImage.	  Eriksson	  PS,	  Perfilieva	  E,	  Björk-­‐Eriksson	  T,	  Alborn	  A-­‐M,	  Nordborg	  C,	  Peterson	  DA,	  Gage	  FH	  (1998)	  Neurogenesis	  in	  the	  adult	  human	  hippocampus.	  Nature	  medicine	  4:1313-­‐1317.	  Feindel	  W,	  Penfield	  W	  (1954)	  Localization	  of	  discharge	  in	  temporal	  lobe	  automatism.	  AMA	  Archives	  of	  Neurology	  &	  Psychiatry	  72:605-­‐630.	  	   175	  Fenwick	  P,	  Schatzman	  M,	  Worsley	  A,	  Adams	  J,	  Stone	  S,	  Baker	  A	  (1984)	  Lucid	  dreaming:	  correspondence	  between	  dreamed	  and	  actual	  events	  in	  one	  subject	  during	  REM	  sleep.	  Biological	  Psychology	  18:243-­‐252.	  Ferguson	  SM,	  Rayport	  M,	  Gardner	  R,	  Kass	  W,	  Weiner	  H,	  Reiser	  MF	  (1969)	  Similarities	  in	  mental	  content	  of	  psychotic	  states,	  spontaneous	  seizures,	  dreams,	  and	  responses	  to	  electrical	  brain	  stimulation	  in	  patients	  with	  temporal	  lobe	  epilepsy.	  Psychosomatic	  Medicine	  31:479-­‐498.	  Fish	  D,	  Gloor	  P,	  Quesney	  F,	  Oliver	  A	  (1993)	  Clinical	  responses	  to	  electrical	  brain	  stimulation	  of	  the	  temporal	  and	  frontal	  lobes	  in	  patients	  with	  epilepsy	  Pathophysiological	  implications.	  Brain	  116:397-­‐414.	  Fisher	  RA	  (1915)	  Frequency	  distribution	  of	  the	  values	  of	  the	  correlation	  coefficient	  in	  samples	  from	  an	  indefinitely	  large	  population.	  Biometrika	  10:507-­‐521.	  Foster	  BL,	  Kaveh	  A,	  Dastjerdi	  M,	  Miller	  KJ,	  Parvizi	  J	  (2013)	  Human	  retrosplenial	  cortex	  displays	  transient	  theta	  phase	  locking	  with	  medial	  temporal	  cortex	  prior	  to	  activation	  during	  autobiographical	  memory	  retrieval.	  The	  Journal	  of	  Neuroscience	  33:10439-­‐10446.	  Foster	  DJ,	  Wilson	  MA	  (2006)	  Reverse	  replay	  of	  behavioural	  sequences	  in	  hippocampal	  place	  cells	  during	  the	  awake	  state.	  Nature	  440:680-­‐683.	  Foulkes	  D,	  Spear	  PS,	  Symonds	  JD	  (1966)	  Individual	  differences	  in	  mental	  activity	  at	  sleep	  onset.	  Journal	  of	  abnormal	  psychology	  71:280.	  Foulkes	  D,	  Vogel	  G	  (1965)	  Mental	  activity	  at	  sleep	  onset.	  Journal	  of	  abnormal	  psychology	  70:231.	  	   176	  Fox	  KCR,	  Andrews-­‐Hanna	  JR,	  Christoff	  K	  (under	  review)	  The	  neurobiology	  of	  self-­‐generated	  thought	  from	  cells	  to	  systems:	  Integrating	  evidence	  from	  lesion	  studies,	  human	  intracranial	  electrophysiology,	  neurochemistry,	  and	  neuroendocrinology.	  Neuroscience.	  Fox	  KCR,	  Andrews-­‐Hanna	  JR,	  Markovic	  J,	  Thompson	  E,	  Christoff	  K	  (in	  preparation)	  The	  wandering	  heart:	  A	  review	  of	  the	  subjective	  diversity	  and	  neural	  correlates	  of	  affect	  in	  mind-­‐wandering	  and	  related	  forms	  of	  self-­‐generated	  thought.	  Emotion.	  Fox	  KCR,	  Christoff	  K	  (2014)	  Metacognitive	  Facilitation	  of	  Spontaneous	  Thought	  Processes:	  When	  Metacognition	  Helps	  the	  Wandering	  Mind	  Find	  Its	  Way.	  In:	  The	  Cognitive	  Neuroscience	  of	  Metacognition,	  pp	  293-­‐319:	  Springer.	  Fox	  KCR,	  Christoff	  K	  (2015)	  Transcranial	  direct	  current	  stimulation	  to	  lateral	  prefrontal	  cortex	  could	  increase	  meta-­‐awareness	  of	  mind	  wandering.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  112:E2414.	  Fox	  KCR,	  Kang	  Y,	  Lifshitz	  M,	  Christoff	  K	  (2016)	  Increasing	  cognitive-­‐emotional	  flexibility	  with	  meditation	  and	  hypnosis:	  The	  cognitive	  neuroscience	  of	  de-­‐automatization.	  In:	  Hypnosis	  and	  Meditation:	  Toward	  an	  Integrative	  Science	  of	  Conscious	  Planes(Raz,	  A.	  and	  Lifshitz,	  M.,	  eds),	  pp	  191-­‐219	  New	  York:	  Oxford	  University	  Press.	  Fox	  KCR,	  Nijeboer	  S,	  Solomonova	  E,	  Domhoff	  GW,	  Christoff	  K	  (2013)	  Dreaming	  as	  mind	  wandering:	  Evidence	  from	  functional	  neuroimaging	  and	  first-­‐person	  content	  reports.	  Front	  Hum	  Neurosci	  7:412.	  Fox	  KCR,	  Spreng	  RN,	  Ellamil	  M,	  Andrews-­‐Hanna	  JR,	  Christoff	  K	  (2015)	  The	  wandering	  brain:	  Meta-­‐analysis	  of	  functional	  neuroimaging	  studies	  of	  mind-­‐wandering	  and	  related	  spontaneous	  thought	  processes.	  NeuroImage	  111:611-­‐621.	  	   177	  Fox	  KCR,	  Thompson	  E,	  Andrews-­‐Hanna	  JR,	  Christoff	  K	  (2014)	  Is	  thinking	  really	  aversive?	  A	  commentary	  on	  Wilson	  et	  al.'s	  "Just	  think:	  The	  challenges	  of	  the	  disengaged	  mind".	  Frontiers	  in	  Psychology:	  Cognition.	  Frank	  J	  (1946)	  Clinical	  survey	  and	  results	  of	  200	  cases	  of	  prefrontal	  leucotomy.	  The	  British	  Journal	  of	  Psychiatry	  92:497-­‐508.	  Frank	  J	  (1950)	  Some	  aspects	  of	  lobotomy	  (prefrontal	  leucotomy)	  under	  psychoanalytic	  scrutiny.	  Psychiatry:	  Journal	  for	  the	  Study	  of	  Interpersonal	  Processes.	  Fransson	  P	  (2006)	  How	  default	  is	  the	  default	  mode	  of	  brain	  function?	  Further	  evidence	  from	  intrinsic	  BOLD	  signal	  fluctuations.	  Neuropsychologia	  44:2836-­‐2845.	  Friston	  KJ,	  Jezzard	  P,	  Turner	  R	  (1994)	  Analysis	  of	  functional	  MRI	  time‐series.	  Human	  Brain	  Mapping	  1:153-­‐171.	  Fuchs	  E,	  Gould	  E	  (2000)	  In	  vivo	  neurogenesis	  in	  the	  adult	  brain:	  regulation	  and	  functional	  implications.	  European	  Journal	  of	  Neuroscience	  12:2211-­‐2214.	  Gainotti	  G	  (2000)	  What	  the	  locus	  of	  brain	  lesion	  tells	  us	  about	  the	  nature	  of	  the	  cognitive	  defect	  underlying	  category-­‐specific	  disorders:	  a	  review.	  Cortex	  36:539-­‐559.	  Gardini	  S,	  Cloninger	  CR,	  Venneri	  A	  (2009)	  Individual	  differences	  in	  personality	  traits	  reflect	  structural	  variance	  in	  specific	  brain	  regions.	  Brain	  Research	  Bulletin	  79:265-­‐270.	  Gelbard-­‐Sagiv	  H,	  Mukamel	  R,	  Harel	  M,	  Malach	  R,	  Fried	  I	  (2008)	  Internally	  generated	  reactivation	  of	  single	  neurons	  in	  human	  hippocampus	  during	  free	  recall.	  Science	  322:96-­‐101.	  Genovese	  CR,	  Lazar	  NA,	  Nichols	  T	  (2002)	  Thresholding	  of	  statistical	  maps	  in	  functional	  neuroimaging	  using	  the	  false	  discovery	  rate.	  NeuroImage	  15:870-­‐878.	  	   178	  Gerlach	  KD,	  Spreng	  RN,	  Madore	  KP,	  Schacter	  DL	  (2014)	  Future	  planning:	  Default	  network	  activity	  couples	  with	  frontoparietal	  control	  network	  and	  reward-­‐processing	  regions	  during	  process	  and	  outcome	  simulations.	  Social	  cognitive	  and	  affective	  neuroscience	  9:1942-­‐1951.	  Giambra	  LM	  (1974)	  Daydreaming	  across	  the	  life	  span:	  Late	  adolescent	  to	  senior	  citizen.	  The	  International	  Journal	  of	  Aging	  and	  Human	  Development	  5:115-­‐140.	  Giambra	  LM	  (1978)	  Adult	  male	  daydreaming	  across	  the	  life	  span:	  A	  replication,	  further	  analyses,	  and	  tentative	  norms	  based	  upon	  retrospective	  reports.	  The	  International	  Journal	  of	  Aging	  and	  Human	  Development	  8:197-­‐228.	  Giambra	  LM	  (1980)	  A	  factor	  analysis	  of	  the	  items	  of	  the	  imaginal	  processes	  inventory.	  Journal	  of	  Clinical	  Psychology	  36:383-­‐409.	  Gillespie	  G	  (1988)	  Lucid	  dreams	  in	  Tibetan	  Buddhism.	  In:	  Conscious	  Mind,	  Sleeping	  Brain(Gackenbach,	  J.	  and	  LaBerge,	  S.	  P.,	  eds),	  pp	  27-­‐66	  New	  York:	  Plenum	  Press.	  Glass	  GV,	  Hopkins	  KD	  (1970)	  Statistical	  methods	  in	  education	  and	  psychology:	  Prentice-­‐Hall	  Englewood	  Cliffs,	  NJ.	  Gloor	  P	  (1990)	  Experiential	  phenomena	  of	  temporal	  lobe	  epilepsy:	  Facts	  and	  hypotheses.	  Brain	  113:1673-­‐1694.	  Gloor	  P,	  Olivier	  A,	  Quesney	  LF,	  Andermann	  F,	  Horowitz	  S	  (1982)	  The	  role	  of	  the	  limbic	  system	  in	  experiential	  phenomena	  of	  temporal	  lobe	  epilepsy.	  Annals	  of	  Neurology	  12:129-­‐144.	  Goldman	  SA,	  Nottebohm	  F	  (1983)	  Neuronal	  production,	  migration,	  and	  differentiation	  in	  a	  vocal	  control	  nucleus	  of	  the	  adult	  female	  canary	  brain.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  80:2390-­‐2394.	  	   179	  Good	  CD,	  Johnsrude	  IS,	  Ashburner	  J,	  Henson	  RN,	  Fristen	  K,	  Frackowiak	  RS	  (2002)	  A	  voxel-­‐based	  morphometric	  study	  of	  ageing	  in	  465	  normal	  adult	  human	  brains.	  In:	  Biomedical	  Imaging,	  2002	  5th	  IEEE	  EMBS	  International	  Summer	  School	  on,	  p	  16	  pp.:	  IEEE.	  Goodenough	  DR,	  Lewis	  HB,	  Shapiro	  A,	  Jaret	  L,	  Sleser	  I	  (1965)	  Dream	  reporting	  following	  abrupt	  and	  gradual	  awakenings	  from	  different	  types	  of	  sleep.	  Journal	  of	  personality	  and	  social	  psychology	  2:170.	  Gorgolewski	  KJ,	  Lurie	  D,	  Urchs	  S,	  Kipping	  JA,	  Craddock	  RC,	  Milham	  MP,	  Margulies	  DS,	  Smallwood	  J	  (2014)	  A	  correspondence	  between	  individual	  differences	  in	  the	  brain's	  intrinsic	  functional	  architecture	  and	  the	  content	  and	  form	  of	  self-­‐generated	  thoughts.	  PLoS	  ONE	  9:e97176.	  Gottesmann	  C	  (1999)	  Neurophysiological	  support	  of	  consciousness	  during	  waking	  and	  sleep.	  Progress	  in	  neurobiology	  59:469-­‐508.	  Green	  CE	  (1968)	  Lucid	  dreams.	  London:	  Hamish	  Hamilton.	  Greenberg	  R,	  Pearlman	  C,	  Brooks	  R,	  Mayer	  R,	  Hartmann	  E	  (1968)	  Dreaming	  and	  Korsakoff's	  psychosis.	  Archives	  of	  General	  Psychiatry	  18:203-­‐209.	  Greicius	  MD,	  Krasnow	  B,	  Reiss	  AL,	  Menon	  V	  (2003)	  Functional	  connectivity	  in	  the	  resting	  brain:	  a	  network	  analysis	  of	  the	  default	  mode	  hypothesis.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  100:253-­‐258.	  Greve	  DN,	  Fischl	  B	  (2009)	  Accurate	  and	  robust	  brain	  image	  alignment	  using	  boundary-­‐based	  registration.	  NeuroImage	  48:63-­‐72.	  Gunaratana	  BH	  (2011)	  Mindfulness	  in	  plain	  English.	  Somerville,	  MA:	  Wisdom	  Publications.	  Guntrip	  H	  (2008)	  Psychoanalytic	  theory,	  therapy,	  and	  the	  self:	  Basic	  Books.	  	   180	  Gupta	  AS,	  van	  der	  Meer	  MA,	  Touretzky	  DS,	  Redish	  AD	  (2010)	  Hippocampal	  replay	  is	  not	  a	  simple	  function	  of	  experience.	  Neuron	  65:695-­‐705.	  Gusnard	  DA,	  Akbudak	  E,	  Shulman	  GL,	  Raichle	  ME	  (2001)	  Medial	  prefrontal	  cortex	  and	  self-­‐referential	  mental	  activity:	  relation	  to	  a	  default	  mode	  of	  brain	  function.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  of	  the	  United	  States	  of	  America	  98:4259-­‐4264.	  Hach	  S,	  Tippett	  LJ,	  Addis	  DR	  (2014)	  Neural	  changes	  associated	  with	  the	  generation	  of	  specific	  past	  and	  future	  events	  in	  depression.	  Neuropsychologia	  65:41-­‐55.	  Haines	  DE	  (2004)	  Neuroanatomy:	  An	  atlas	  of	  structures,	  sections,	  and	  systems:	  Lippincott	  Williams	  &	  Wilkins.	  Halgren	  E,	  Chauvel	  P	  (1992)	  Experimental	  phenomena	  evoked	  by	  human	  brain	  electrical	  stimulation.	  Advances	  in	  neurology	  63:123-­‐140.	  Halgren	  E,	  Walter	  RD,	  Cherlow	  DG,	  Crandall	  PH	  (1978)	  Mental	  phenomena	  evoked	  by	  electrical	  stimulation	  of	  the	  human	  hippocampal	  formation	  and	  amygdala.	  Brain	  101:83-­‐115.	  Hall	  CS,	  Van	  de	  Castle	  RL	  (1966)	  The	  content	  analysis	  of	  dreams.	  Hamilton	  JP,	  Etkin	  A,	  Furman	  DJ,	  Lemus	  MG,	  Johnson	  RF,	  Gotlib	  IH	  (2012)	  Functional	  neuroimaging	  of	  major	  depressive	  disorder:	  a	  meta-­‐analysis	  and	  new	  integration	  of	  baseline	  activation	  and	  neural	  response	  data.	  American	  Journal	  of	  Psychiatry	  169:693-­‐703.	  Hamilton	  JP,	  Furman	  DJ,	  Chang	  C,	  Thomason	  ME,	  Dennis	  E,	  Gotlib	  IH	  (2011)	  Default-­‐mode	  and	  task-­‐positive	  network	  activity	  in	  major	  depressive	  disorder:	  implications	  for	  adaptive	  and	  maladaptive	  rumination.	  Biological	  psychiatry	  70:327-­‐333.	  	   181	  Hartmann	  E	  (1996)	  We	  do	  not	  dream	  of	  the	  Three	  R's:	  a	  study	  and	  implications.	  Sleep	  Research	  25:136.	  Hasenkamp	  W,	  Barsalou	  LW	  (2012)	  Effects	  of	  meditation	  experience	  on	  functional	  connectivity	  of	  distributed	  brain	  networks.	  Front	  Hum	  Neurosci	  6:38.	  Hasenkamp	  W,	  Wilson-­‐Mendenhall	  CD,	  Duncan	  E,	  Barsalou	  LW	  (2012)	  Mind	  wandering	  and	  attention	  during	  focused	  meditation:	  a	  fine-­‐grained	  temporal	  analysis	  of	  fluctuating	  cognitive	  states.	  NeuroImage	  59:750-­‐760.	  Hassabis	  D,	  Kumaran	  D,	  Vann	  SD,	  Maguire	  EA	  (2007)	  Patients	  with	  hippocampal	  amnesia	  cannot	  imagine	  new	  experiences.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  104:1726-­‐1731.	  Hawkinson	  JE,	  Ross	  AJ,	  Parthasarathy	  S,	  Scott	  DJ,	  Laramee	  EA,	  Posecion	  LJ,	  Rekshan	  WR,	  Sheau	  KE,	  Njaka	  ND,	  Bayley	  PJ	  (2012)	  Quantification	  of	  adverse	  events	  associated	  with	  functional	  MRI	  scanning	  and	  with	  real-­‐time	  fMRI-­‐based	  training.	  International	  journal	  of	  behavioral	  medicine	  19:372-­‐381.	  Haxby	  JV	  (2012)	  Multivariate	  pattern	  analysis	  of	  fMRI:	  the	  early	  beginnings.	  NeuroImage	  62:852-­‐855.	  Hayashi	  M,	  Katoh	  K,	  Hori	  T	  (1999)	  Hypnagogic	  imagery	  and	  EEG	  activity.	  Perceptual	  and	  motor	  skills	  88:676-­‐678.	  Heiss	  W-­‐D,	  Pawlik	  G,	  Herholz	  K,	  Wagner	  R,	  Wienhard	  K	  (1985)	  Regional	  cerebral	  glucose	  metabolism	  in	  man	  during	  wakefulness,	  sleep,	  and	  dreaming.	  Brain	  Research	  327:362-­‐366.	  Hobson	  JA	  (2009)	  REM	  sleep	  and	  dreaming:	  towards	  a	  theory	  of	  protoconsciousness.	  Nat	  Rev	  Neurosci	  10:803-­‐813.	  	   182	  Hobson	  JA,	  Pace-­‐Schott	  EF,	  Stickgold	  R	  (2000)	  Dreaming	  and	  the	  brain:	  Toward	  a	  cognitive	  neuroscience	  of	  conscious	  states.	  Behavioral	  and	  Brain	  Sciences	  23:793-­‐842.	  Holzinger	  B,	  LaBerge	  SP,	  Levitan	  L	  (2006)	  Psychophysiological	  correlates	  of	  lucid	  dreaming.	  Dreaming	  16:88.	  Hori	  T,	  Hayashi	  M,	  Morikawa	  T	  (1994)	  Topographical	  EEG	  changes	  and	  the	  hypnagogic	  experience.	  Horowitz	  MJ,	  Adams	  JE,	  Rutkin	  BB	  (1968)	  Visual	  imagery	  on	  brain	  stimulation.	  Archives	  of	  General	  Psychiatry	  19:469-­‐486.	  Houk	  JC,	  Wise	  SP	  (1995)	  Distributed	  modular	  architectures	  linking	  basal	  ganglia,	  cerebellum,	  and	  cerebral	  cortex:	  their	  role	  in	  planning	  and	  controlling	  action.	  Cerebral	  Cortex	  5:95-­‐110.	  Hua	  K,	  Zhang	  J,	  Wakana	  S,	  Jiang	  H,	  Li	  X,	  Reich	  DS,	  Calabresi	  PA,	  Pekar	  JJ,	  van	  Zijl	  PC,	  Mori	  S	  (2008)	  Tract	  probability	  maps	  in	  stereotaxic	  spaces:	  analyses	  of	  white	  matter	  anatomy	  and	  tract-­‐specific	  quantification.	  NeuroImage	  39:336-­‐347.	  Huettel	  SA,	  Song	  AW,	  McCarthy	  G	  (2004)	  Functional	  magnetic	  resonance	  imaging:	  Sinauer	  Associates	  Sunderland.	  Hume	  RE	  (1921)	  The	  thirteen	  principal	  Upanishads:	  Translated	  from	  the	  sanskrit	  with	  an	  outline	  of	  the	  philosophy	  of	  the	  Upanishads	  and	  an	  annotated	  bibliography:	  H.	  Milford,	  Oxford	  University	  Press.	  Irish	  M,	  Piguet	  O,	  Hodges	  JR	  (2012)	  Self-­‐projection	  and	  the	  default	  network	  in	  frontotemporal	  dementia.	  Nature	  Reviews	  Neurology	  8:152-­‐161.	  Jacobs	  J,	  Lega	  B,	  Anderson	  C	  (2012)	  Explaining	  how	  brain	  stimulation	  can	  evoke	  memories.	  Journal	  of	  Cognitive	  Neuroscience	  24:553-­‐563.	  	   183	  James	  W	  (1985)	  The	  varieties	  of	  religious	  experience:	  Harvard	  University	  Press.	  Jaynes	  J	  (1976)	  The	  origin	  of	  consciousness	  in	  the	  breakdown	  of	  the	  bicameral	  mind:	  Houghton	  Mifflin	  Harcourt.	  Jenkinson	  M,	  Bannister	  P,	  Brady	  M,	  Smith	  S	  (2002)	  Improved	  optimization	  for	  the	  robust	  and	  accurate	  linear	  registration	  and	  motion	  correction	  of	  brain	  images.	  NeuroImage	  17:825-­‐841.	  Jenkinson	  M,	  Beckmann	  CF,	  Behrens	  TE,	  Woolrich	  MW,	  Smith	  SM	  (2012)	  FSL.	  NeuroImage	  62:782-­‐790.	  Jenkinson	  M,	  Smith	  S	  (2001)	  A	  global	  optimisation	  method	  for	  robust	  affine	  registration	  of	  brain	  images.	  Medical	  image	  analysis	  5:143-­‐156.	  Ji	  D,	  Wilson	  MA	  (2007)	  Coordinated	  memory	  replay	  in	  the	  visual	  cortex	  and	  hippocampus	  during	  sleep.	  Nature	  Neuroscience	  10:100-­‐107.	  Jones	  B	  (1991)	  Paradoxical	  sleep	  and	  its	  chemical/structural	  substrates	  in	  the	  brain.	  Neuroscience	  40:637-­‐656.	  Jones	  B	  (2005)	  From	  waking	  to	  sleeping:	  neuronal	  and	  chemical	  substrates.	  Trends	  in	  pharmacological	  sciences	  26:578-­‐586.	  Jones	  DK,	  Christiansen	  K,	  Chapman	  R,	  Aggleton	  J	  (2013)	  Distinct	  subdivisions	  of	  the	  cingulum	  bundle	  revealed	  by	  diffusion	  MRI	  fibre	  tracking:	  implications	  for	  neuropsychological	  investigations.	  Neuropsychologia	  51:67-­‐78.	  Kaasinen	  V,	  Maguire	  R,	  Kurki	  T,	  Brück	  A,	  Rinne	  J	  (2005)	  Mapping	  brain	  structure	  and	  personality	  in	  late	  adulthood.	  NeuroImage	  24:315-­‐322.	  Kabat-­‐Zinn	  J	  (2003)	  Mindfulness-­‐based	  stress	  reduction	  (MBSR).	  Constructivism	  in	  the	  Human	  Sciences.	  	   184	  Kahane	  P,	  Hoffmann	  D,	  Minotti	  L,	  Berthoz	  A	  (2003)	  Reappraisal	  of	  the	  human	  vestibular	  cortex	  by	  cortical	  electrical	  stimulation	  study.	  Annals	  of	  Neurology	  54:615-­‐624.	  Kahn	  D,	  Hobson	  JA	  (2005)	  State-­‐dependent	  thinking:	  A	  comparison	  of	  waking	  and	  dreaming	  thought.	  Consciousness	  and	  cognition	  14:429-­‐438.	  Kahn	  D,	  Pace-­‐Schott	  EF,	  Hobson	  JA	  (1997)	  Consciousness	  in	  waking	  and	  dreaming:	  the	  roles	  of	  neuronal	  oscillation	  and	  neuromodulation	  in	  determining	  similarities	  and	  differences.	  Neuroscience	  78:13-­‐38.	  Kane	  MJ,	  Brown	  LH,	  McVay	  JC,	  Silvia	  PJ,	  Myin-­‐Germeys	  I,	  Kwapil	  TR	  (2007)	  For	  whom	  the	  mind	  wanders,	  and	  when:	  an	  experience-­‐sampling	  study	  of	  working	  memory	  and	  executive	  control	  in	  daily	  life.	  Psychol	  Sci	  18:614-­‐621.	  Kier	  EL,	  Staib	  LH,	  Davis	  LM,	  Bronen	  RA	  (2004)	  MR	  imaging	  of	  the	  temporal	  stem:	  anatomic	  dissection	  tractography	  of	  the	  uncinate	  fasciculus,	  inferior	  occipitofrontal	  fasciculus,	  and	  Meyer’s	  loop	  of	  the	  optic	  radiation.	  American	  Journal	  of	  Neuroradiology	  25:677-­‐691.	  Killingsworth	  MA,	  Gilbert	  DT	  (2010)	  A	  wandering	  mind	  is	  an	  unhappy	  mind.	  Science	  330:932.	  Kirchoff	  BA,	  Schapiro	  ML,	  Buckner	  RL	  (2005)	  Orthographic	  distinctiveness	  and	  semantic	  elaboration	  provide	  separate	  contributions	  to	  memory.	  Journal	  of	  Cognitive	  Neuroscience	  17:1841-­‐1854.	  Klein	  SB,	  Loftus	  J,	  Kihlstrom	  JF	  (2002)	  Memory	  and	  temporal	  experience:	  the	  effects	  of	  episodic	  memory	  loss	  on	  an	  amnesic	  patient's	  ability	  to	  remember	  the	  past	  and	  imagine	  the	  future.	  Social	  Cognition	  20:353-­‐379.	  	   185	  Klinger	  E	  (1978)	  Modes	  of	  normal	  conscious	  flow.	  In:	  The	  stream	  of	  consciousness,	  pp	  225-­‐258:	  Springer.	  Klinger	  E	  (1987)	  Current	  concerns	  and	  disengagement	  from	  incentives.	  In:	  Motivation,	  intention,	  and	  volition,	  pp	  337-­‐347:	  Springer.	  Klinger	  E	  (2008)	  Daydreaming	  and	  Fantasizing:	  Thought	  Flow	  and	  Motivation.	  In:	  Handbook	  of	  Imagination	  and	  Mental	  Simulation(Markman,	  K.	  D.	  et	  al.,	  eds),	  pp	  225-­‐239	  New	  York:	  Psychology	  Press.	  Klinger	  E	  (2013)	  Goal	  Commitments	  and	  the	  content	  of	  thoughts	  and	  dreams:	  basic	  principles.	  Frontiers	  in	  Psychology	  4.	  Klinger	  E,	  Cox	  WM	  (1987)	  Dimensions	  of	  thought	  flow	  in	  everyday	  life.	  Imagination,	  Cognition	  and	  Personality	  7:105-­‐128.	  Klinger	  E,	  Cox	  WM	  (2004)	  Motivation	  and	  the	  theory	  of	  current	  concerns.	  Handbook	  of	  motivational	  counseling:	  Concepts,	  approaches,	  and	  assessment	  3-­‐27.	  Klinger	  E,	  Cox	  WM	  (2011)	  Motivation	  and	  the	  goal	  theory	  of	  current	  concerns.	  Handbook	  of	  motivational	  counseling:	  Goal-­‐based	  approaches	  to	  assessment	  and	  intervention	  with	  addiction	  and	  other	  problems	  1-­‐47.	  Klinger	  E,	  Kroll-­‐Mensing	  D	  (1995)	  Idiothetic	  assessment	  experience	  sampling	  and	  motivational	  analysis.	  Clinical	  personality	  assessment:	  Practical	  approaches	  267-­‐277.	  Knierim	  JJ	  (2009)	  Imagining	  the	  possibilities:	  ripples,	  routes,	  and	  reactivation.	  Neuron	  63:421-­‐423.	  Koenigs	  M,	  Tranel	  D,	  Damasio	  AR	  (2007)	  The	  lesion	  method	  in	  cognitive	  neuroscience.	  Handbook	  of	  psychophysiology	  139-­‐158.	  	   186	  Kohut	  H	  (2013)	  The	  analysis	  of	  the	  self:	  A	  systematic	  approach	  to	  the	  psychoanalytic	  treatment	  of	  narcissistic	  personality	  disorders:	  University	  of	  Chicago	  Press.	  Koolschijn	  P,	  van	  Haren	  NE,	  Lensvelt‐Mulders	  GJ,	  Pol	  H,	  Hilleke	  E,	  Kahn	  RS	  (2009)	  Brain	  volume	  abnormalities	  in	  major	  depressive	  disorder:	  A	  meta‐analysis	  of	  magnetic	  resonance	  imaging	  studies.	  Human	  Brain	  Mapping	  30:3719-­‐3735.	  Krans	  J,	  de	  Bree	  J,	  Moulds	  ML	  (2015)	  Involuntary	  cognitions	  in	  everyday	  life:	  exploration	  of	  type,	  quality,	  content,	  and	  function.	  Frontiers	  in	  psychiatry	  6.	  Krebs	  C	  (2011)	  Neuroscience:	  Lippincott	  Williams	  &	  Wilkins.	  Kroll-­‐Mensing	  D	  (1992)	  Differentiating	  anxiety	  and	  depression:	  An	  experience	  sampling	  analysis.	  Unpublished	  Ph	  D	  dissertation,	  University	  of	  Minnesota.	  Kucyi	  A,	  Salomons	  TV,	  Davis	  KD	  (2013)	  Mind	  wandering	  away	  from	  pain	  dynamically	  engages	  antinociceptive	  and	  default	  mode	  brain	  networks.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  110:18692-­‐18697.	  Kuijsten	  M	  (2006)	  Reflections	  on	  the	  dawn	  of	  consciousness:	  Julian	  Jaynes's	  bicameral	  mind	  theory	  revisited:	  Julian	  Jaynes	  Society.	  Kuiken	  D,	  Dunn	  S,	  LoVerso	  T	  (2008)	  Expressive	  writing	  about	  dreams	  that	  follow	  trauma	  and	  loss.	  Dreaming	  18:77.	  Kvavilashvili	  L,	  Mandler	  G	  (2004)	  Out	  of	  one’s	  mind:	  A	  study	  of	  involuntary	  semantic	  memories.	  Cognitive	  psychology	  48:47-­‐94.	  Kwan	  D,	  Carson	  N,	  Addis	  D,	  Rosenbaum	  R	  (2010)	  Deficits	  in	  past	  remembering	  extend	  to	  future	  imagining	  in	  a	  case	  of	  developmental	  amnesia.	  Neuropsychologia	  48:3179-­‐3186.	  	   187	  LaBerge	  SP	  (1980)	  Lucid	  dreaming	  as	  a	  learnable	  skill:	  A	  case	  study.	  Perceptual	  and	  motor	  skills	  51:1039-­‐1042.	  LaBerge	  SP,	  Levitan	  L	  (1995)	  Validity	  established	  of	  DreamLight	  cues	  for	  eliciting	  lucid	  dreaming.	  Dreaming	  5:159.	  LaBerge	  SP,	  Levitan	  L,	  Dement	  WC	  (1986)	  Lucid	  dreaming:	  Physiological	  correlates	  of	  consciousness	  during	  REM	  sleep.	  Journal	  of	  Mind	  and	  Behavior.	  LaBerge	  SP,	  Nagel	  LE,	  Dement	  WC,	  Zarcone	  Jr	  VP	  (1981)	  Lucid	  dreaming	  verified	  by	  volitional	  communication	  during	  REM	  sleep.	  Perceptual	  and	  motor	  skills	  52:727-­‐732.	  Laird	  AR,	  Fox	  PM,	  Price	  CJ,	  Glahn	  DC,	  Uecker	  AM,	  Lancaster	  JL,	  Turkeltaub	  PE,	  Kochunov	  P,	  Fox	  PT	  (2005)	  ALE	  meta‐analysis:	  Controlling	  the	  false	  discovery	  rate	  and	  performing	  statistical	  contrasts.	  Human	  Brain	  Mapping	  25:155-­‐164.	  Lee	  H,	  Hong	  S,	  Seo	  D,	  Tae	  W,	  Hong	  S	  (2000)	  Mapping	  of	  functional	  organization	  in	  human	  visual	  cortex	  Electrical	  cortical	  stimulation.	  Neurology	  54:849-­‐854.	  Lena	  I,	  Parrot	  S,	  Deschaux	  O,	  Muffat‐Joly	  S,	  Sauvinet	  V,	  Renaud	  B,	  Suaud‐Chagny	  MF,	  Gottesmann	  C	  (2005)	  Variations	  in	  extracellular	  levels	  of	  dopamine,	  noradrenaline,	  glutamate,	  and	  aspartate	  across	  the	  sleep–wake	  cycle	  in	  the	  medial	  prefrontal	  cortex	  and	  nucleus	  accumbens	  of	  freely	  moving	  rats.	  Journal	  of	  neuroscience	  research	  81:891-­‐899.	  Lerch	  JP,	  Yiu	  AP,	  Martinez-­‐Canabal	  A,	  Pekar	  T,	  Bohbot	  VD,	  Frankland	  PW,	  Henkelman	  RM,	  Josselyn	  SA,	  Sled	  JG	  (2011)	  Maze	  training	  in	  mice	  induces	  MRI-­‐detectable	  brain	  shape	  changes	  specific	  to	  the	  type	  of	  learning.	  NeuroImage	  54:2086-­‐2095.	  	   188	  Lilly	  JC	  (1956)	  Mental	  effects	  of	  reduction	  of	  ordinary	  levels	  of	  physical	  stimuli	  on	  intact,	  healthy	  persons.	  Psychiatric	  Research	  Reports.	  Lilly	  JC	  (1972)	  Programming	  and	  metaprogramming	  in	  the	  human	  biocomputer:	  Julian	  P.	  Lilly	  JC	  (1977)	  The	  deep	  self.	  New	  York:	  Simon	  &	  Schuster.	  Loftus	  EF	  (1996)	  Memory	  distortion	  and	  false	  memory	  creation.	  Journal	  of	  the	  American	  Academy	  of	  Psychiatry	  and	  the	  Law	  Online	  24:281-­‐295.	  Loomis	  AL,	  Harvey	  EN,	  Hobart	  G	  (1935)	  Further	  observations	  on	  the	  potential	  rhythms	  of	  the	  cerebral	  cortex	  during	  sleep.	  Science.	  Loomis	  AL,	  Harvey	  EN,	  Hobart	  G	  (1937)	  Cerebral	  states	  during	  sleep,	  as	  studied	  by	  human	  brain	  potentials.	  Journal	  of	  experimental	  psychology	  21:127.	  Lovden	  M,	  Wenger	  E,	  Martensson	  J,	  Lindenberger	  U,	  Backman	  L	  (2013)	  Structural	  brain	  plasticity	  in	  adult	  learning	  and	  development.	  Neuroscience	  and	  biobehavioral	  reviews.	  Luders	  E,	  Clark	  K,	  Narr	  KL,	  Toga	  AW	  (2011)	  Enhanced	  brain	  connectivity	  in	  long-­‐term	  meditation	  practitioners.	  NeuroImage	  57:1308-­‐1316.	  Luria	  AR	  (1976)	  The	  working	  brain:	  An	  introduction	  to	  neuropsychology:	  Basic	  Books.	  Madsen	  P,	  Schmidt	  J,	  Wildschiodtz	  G,	  Friberg	  L,	  Holm	  S,	  Vorstrup	  S,	  Lassen	  N	  (1991)	  Cerebral	  O2	  metabolism	  and	  cerebral	  blood	  flow	  in	  humans	  during	  deep	  and	  rapid-­‐eye-­‐movement	  sleep.	  Journal	  of	  Applied	  Physiology	  70:2597-­‐2601.	  Mahl	  GF,	  Rothenberg	  A,	  Delgado	  JM,	  Hamlin	  H	  (1964)	  Psychological	  responses	  in	  the	  human	  to	  intracerebral	  electrical	  stimulation.	  Psychosomatic	  Medicine	  26:337-­‐368.	  	   189	  Makris	  N,	  Kennedy	  DN,	  McInerney	  S,	  Sorensen	  AG,	  Wang	  R,	  Caviness	  VS,	  Jr.,	  Pandya	  DN	  (2005)	  Segmentation	  of	  subcomponents	  within	  the	  superior	  longitudinal	  fascicle	  in	  humans:	  a	  quantitative,	  in	  vivo,	  DT-­‐MRI	  study.	  Cereb	  Cortex	  15:854-­‐869.	  Maldjian	  JA,	  Laurienti	  PJ,	  Kraft	  RA,	  Burdette	  JH	  (2003)	  An	  automated	  method	  for	  neuroanatomic	  and	  cytoarchitectonic	  atlas-­‐based	  interrogation	  of	  fMRI	  data	  sets.	  NeuroImage	  19:1233-­‐1239.	  Maquet	  P	  (1995)	  Sleep	  function	  (s)	  and	  cerebral	  metabolism.	  Behavioural	  brain	  research	  69:75-­‐83.	  Maquet	  P	  (2010)	  Understanding	  non	  rapid	  eye	  movement	  sleep	  through	  neuroimaging.	  World	  J	  Biol	  Psychiatry	  11	  Suppl	  1:9-­‐15.	  Maquet	  P,	  Dive	  D,	  Salmon	  E,	  Sadzot	  B,	  Franco	  G,	  Poirrier	  R,	  von	  Frenckell	  R,	  Franck	  G	  (1990)	  Cerebral	  glucose	  utilization	  during	  sleep-­‐wake	  cycle	  in	  man	  determined	  by	  positron	  emission	  tomography	  and	  [18	  F]	  2-­‐fluoro-­‐2-­‐deoxy-­‐d-­‐glucose	  method.	  Brain	  Research	  513:136-­‐143.	  Maquet	  P,	  Péters	  J-­‐M,	  Aerts	  J,	  Delfiore	  G,	  Degueldre	  C,	  Luxen	  A,	  Franck	  G	  (1996)	  Functional	  neuroanatomy	  of	  human	  rapid-­‐eye-­‐movement	  sleep	  and	  dreaming.	  Nature	  383:163-­‐166.	  Mar	  RA,	  Mason	  MF,	  Litvack	  A	  (2012)	  How	  daydreaming	  relates	  to	  life	  satisfaction,	  loneliness,	  and	  social	  support:	  the	  importance	  of	  gender	  and	  daydream	  content.	  Consciousness	  and	  cognition	  21:401-­‐407.	  Mason	  MF,	  Brown	  K,	  Mar	  RA,	  Smallwood	  J	  (2013)	  Driver	  of	  discontent	  or	  escape	  vehicle:	  the	  affective	  consequences	  of	  mindwandering.	  Frontiers	  in	  Psychology	  4.	  	   190	  Mason	  MF,	  Norton	  MI,	  Van	  Horn	  JD,	  Wegner	  DM,	  Grafton	  ST,	  Macrae	  CN	  (2007)	  Wandering	  Minds:	  The	  Default	  Network	  and	  Stimulus-­‐Independent	  Thought.	  Science	  315:393-­‐395.	  Mathern	  GW,	  Miller	  JW	  (2013)	  Outcomes	  for	  temporal	  lobe	  epilepsy	  operations	  may	  not	  be	  equal	  A	  call	  for	  an	  RCT	  of	  ATL	  vs	  SAH.	  Neurology	  80:1630-­‐1631.	  May	  A,	  Gaser	  C	  (2006)	  Magnetic	  resonance-­‐based	  morphometry:	  a	  window	  into	  structural	  plasticity	  of	  the	  brain.	  Current	  opinion	  in	  neurology	  19:407-­‐411.	  Mazoyer	  B,	  Zago	  L,	  Mellet	  E,	  Bricogne	  S,	  Etard	  O,	  Houdé	  O,	  Crivello	  F,	  Joliot	  M,	  Petit	  L,	  Tzourio-­‐Mazoyer	  N	  (2001)	  Cortical	  networks	  for	  working	  memory	  and	  executive	  functions	  sustain	  the	  conscious	  resting	  state	  in	  man.	  Brain	  Research	  Bulletin	  54:287-­‐298.	  McGuire	  PK,	  Paulesu	  E,	  Frackowiak	  RSJ,	  Frith	  CD	  (1996)	  Brain	  activity	  during	  stimulus	  independent	  thought.	  Neuroreport	  7:2095-­‐2099.	  McKiernan	  KA,	  D'Angelo	  BR,	  Kaufman	  JN,	  Binder	  JR	  (2006)	  Interrupting	  the	  "stream	  of	  consciousness":	  an	  fMRI	  investigation.	  NeuroImage	  29:1185-­‐1191.	  McMahon	  D	  (2013)	  Divine	  fury:	  a	  history	  of	  genius:	  Basic	  Books.	  McMillan	  RL,	  Kaufman	  SB,	  Singer	  JL	  (2013)	  Ode	  to	  positive	  constructive	  daydreaming.	  Front	  Psychol	  4:626.	  McVay	  JC,	  Kane	  MJ,	  Kwapil	  TR	  (2009)	  Tracking	  the	  train	  of	  thought	  from	  the	  laboratory	  into	  everyday	  life:	  An	  experience-­‐sampling	  study	  of	  mind	  wandering	  across	  controlled	  and	  ecological	  contexts.	  Psychonomic	  Bulletin	  &	  Review	  16:857-­‐863.	  	   191	  Mechelli	  A,	  Price	  CJ,	  Friston	  KJ,	  Ashburner	  J	  (2005)	  Voxel-­‐based	  morphometry	  of	  the	  human	  brain:	  methods	  and	  applications.	  Current	  Medical	  Imaging	  Reviews	  1:105-­‐113.	  Middleton	  FA,	  Strick	  PL	  (1994)	  Anatomical	  evidence	  for	  cerebellar	  and	  basal	  ganglia	  involvement	  in	  higher	  cognitive	  function.	  Science	  266:458-­‐461.	  Milner	  B,	  Corkin	  S,	  Teuber	  H-­‐L	  (1968)	  Further	  analysis	  of	  the	  hippocampal	  amnesic	  syndrome:	  14-­‐year	  follow-­‐up	  study	  of	  HM.	  Neuropsychologia	  6:215-­‐234.	  Monroe	  LJ,	  Rechtschaffen	  A,	  Foulkes	  D,	  Jensen	  J	  (1965)	  Discriminability	  of	  REM	  and	  NREM	  reports.	  Journal	  of	  personality	  and	  social	  psychology	  2:456.	  Monti	  JM,	  Monti	  D	  (2007)	  The	  involvement	  of	  dopamine	  in	  the	  modulation	  of	  sleep	  and	  waking.	  Sleep	  medicine	  reviews	  11:113-­‐133.	  Morris	  HH,	  Luders	  H,	  Lesser	  RP,	  Dinner	  DS,	  Hahn	  J	  (1984)	  Transient	  neuropsychological	  abnormalities	  (including	  Gerstmann's	  Symdrome)	  during	  cortical	  stimulation.	  Neurology	  34:877-­‐877.	  Moscovitch	  M	  (1992)	  Memory	  and	  working-­‐with-­‐memory:	  A	  component	  process	  model	  based	  on	  modules	  and	  central	  systems.	  Journal	  of	  Cognitive	  Neuroscience	  4:257-­‐267.	  Moss	  J,	  Schunn	  CD,	  Schneider	  W,	  McNamara	  DS	  (2013)	  The	  nature	  of	  mind	  wandering	  during	  reading	  varies	  with	  the	  cognitive	  control	  demands	  of	  the	  reading	  strategy.	  Brain	  Research	  1539:48-­‐60.	  Mrazek	  MD,	  Franklin	  MS,	  Phillips	  DT,	  Baird	  B,	  Schooler	  JW	  (2013)	  Mindfulness	  training	  improves	  working	  memory	  capacity	  and	  GRE	  performance	  while	  reducing	  mind	  wandering.	  Psychol	  Sci	  24:776-­‐781.	  	   192	  Mulak	  A,	  Kahane	  P,	  Hoffmann	  D,	  Minotti	  L,	  Bonaz	  B	  (2008)	  Brain	  mapping	  of	  digestive	  sensations	  elicited	  by	  cortical	  electrical	  stimulations.	  Neurogastroenterology	  &	  Motility	  20:588-­‐596.	  Mullan	  S,	  Penfield	  W	  (1959)	  Illusions	  of	  comparative	  interpretation	  and	  emotion:	  production	  by	  epileptic	  discharge	  and	  by	  electrical	  stimulation	  in	  the	  temporal	  cortex.	  AMA	  Archives	  of	  Neurology	  &	  Psychiatry	  81:269-­‐284.	  Müller	  N,	  Knight	  R	  (2006)	  The	  functional	  neuroanatomy	  of	  working	  memory:	  contributions	  of	  human	  brain	  lesion	  studies.	  Neuroscience	  139:51-­‐58.	  Murphy	  K,	  Garavan	  H	  (2005)	  Deriving	  the	  optimal	  number	  of	  events	  for	  an	  event-­‐related	  fMRI	  study	  based	  on	  the	  spatial	  extent	  of	  activation.	  NeuroImage	  27:771-­‐777.	  Nielsen	  TA	  (1992)	  A	  self-­‐observational	  study	  of	  spontaneous	  hypnagogic	  imagery	  using	  the	  upright	  napping	  procedure.	  Imagination,	  Cognition	  and	  Personality	  11:353-­‐366.	  Nielsen	  TA	  (1999)	  Mentation	  during	  sleep:	  the	  NREM/REM	  distinction.	  Handbook	  of	  behavioral	  state	  control	  Cellular	  and	  molecular	  mechanisms	  101-­‐129.	  Nielsen	  TA	  (2000)	  A	  review	  of	  mentation	  in	  REM	  and	  NREM	  sleep:“covert”	  REM	  sleep	  as	  a	  possible	  reconciliation	  of	  two	  opposing	  models.	  Behavioral	  and	  Brain	  Sciences	  23:851-­‐866.	  Nielsen	  TA	  (2014)	  What	  Is	  the	  Current	  Status	  of	  Your	  “Covert	  REM	  Process”	  Theory,	  Especially	  in	  the	  Light	  of	  the	  New	  Protoconsciousness	  Hypothesis?	  In:	  Dream	  Consciousness,	  pp	  175-­‐180:	  Springer.	  Niendam	  TA,	  Laird	  AR,	  Ray	  KL,	  Dean	  YM,	  Glahn	  DC,	  Carter	  CS	  (2012)	  Meta-­‐analytic	  evidence	  for	  a	  superordinate	  cognitive	  control	  network	  subserving	  diverse	  executive	  functions.	  Cognitive,	  Affective,	  &	  Behavioral	  Neuroscience	  12:241-­‐268.	  	   193	  Nir	  Y,	  Tononi	  G	  (2010)	  Dreaming	  and	  the	  brain:	  from	  phenomenology	  to	  neurophysiology.	  Trends	  Cogn	  Sci	  14:88-­‐100.	  Norman	  KA,	  Polyn	  SM,	  Detre	  GJ,	  Haxby	  JV	  (2006)	  Beyond	  mind-­‐reading:	  multi-­‐voxel	  pattern	  analysis	  of	  fMRI	  data.	  Trends	  in	  Cognitive	  Sciences	  10:424-­‐430.	  Olejniczak	  P	  (2006)	  Neurophysiologic	  basis	  of	  EEG.	  Journal	  of	  clinical	  neurophysiology	  23:186-­‐189.	  Olson	  IR,	  Plotzker	  A,	  Ezzyat	  Y	  (2007)	  The	  enigmatic	  temporal	  pole:	  a	  review	  of	  findings	  on	  social	  and	  emotional	  processing.	  Brain	  130:1718-­‐1731.	  Ostrowsky	  K,	  Desestret	  V,	  Ryvlin	  P,	  Coste	  S,	  Mauguière	  F	  (2002)	  Direct	  electrical	  stimulations	  of	  the	  temporal	  pole	  in	  human.	  Epileptic	  disorders:	  international	  epilepsy	  journal	  with	  videotape	  4:S23-­‐27.	  Ouspensky	  PD	  (1957)	  The	  fourth	  way:	  Routledge	  &	  Kegan	  Paul.	  Pace-­‐Schott	  EF	  (2007)	  The	  frontal	  lobes	  and	  dreaming.	  Pace-­‐Schott	  EF,	  Hobson	  JA	  (2002)	  The	  neurobiology	  of	  sleep:	  genetics,	  cellular	  physiology	  and	  subcortical	  networks.	  Nature	  Reviews	  Neuroscience	  3:591-­‐605.	  Parvizi	  J	  (2009)	  Corticocentric	  myopia:	  old	  bias	  in	  new	  cognitive	  sciences.	  Trends	  in	  Cognitive	  Sciences	  13:354-­‐359.	  Pausanius	  (1964)	  Description	  of	  Greece:	  Books	  I	  and	  II:	  Loeb	  Classical	  Library.	  Penfield	  W	  (1958)	  Some	  mechanisms	  of	  consciousness	  discovered	  during	  electrical	  stimulation	  of	  the	  brain.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  of	  the	  United	  States	  of	  America	  44:51.	  	   194	  Penfield	  W,	  Boldrey	  E	  (1937)	  Somatic	  motor	  and	  sensory	  representation	  in	  the	  cerebral	  cortex	  of	  man	  as	  studied	  by	  electrical	  stimulation.	  Brain:	  A	  journal	  of	  neurology	  60:389-­‐443.	  Penfield	  W,	  Perot	  P	  (1963)	  The	  brain’s	  record	  of	  auditory	  and	  visual	  experience	  A	  final	  summary	  and	  discussion.	  Brain	  86:595-­‐696.	  Phan	  KL,	  Wager	  T,	  Taylor	  SF,	  Liberzon	  I	  (2002)	  Functional	  neuroanatomy	  of	  emotion:	  a	  meta-­‐analysis	  of	  emotion	  activation	  studies	  in	  PET	  and	  fMRI.	  NeuroImage	  16:331-­‐348.	  Philippi	  CL,	  Tranel	  D,	  Duff	  M,	  Rudrauf	  D	  (2014)	  Damage	  to	  the	  default	  mode	  network	  disrupts	  autobiographical	  memory	  retrieval.	  Social	  cognitive	  and	  affective	  neuroscience	  nsu070.	  Piehler	  R	  (1950)	  Ueber	  das	  Traumleben	  Leukotomierter	  (Vorlaeufige	  Mitteilung).	  Nervenarzt	  21:517-­‐521.	  Poerio	  GL,	  Totterdell	  P,	  Miles	  E	  (2013)	  Mind-­‐wandering	  and	  negative	  mood:	  Does	  one	  thing	  really	  lead	  to	  another?	  Consciousness	  and	  cognition	  22:1412-­‐1421.	  Poldrack	  R	  (2006)	  Can	  cognitive	  processes	  be	  inferred	  from	  neuroimaging	  data?	  Trends	  in	  Cognitive	  Sciences	  10:59-­‐63.	  Prabhavananda	  S,	  Manchester	  F,	  Isherwood	  C	  (1984)	  The	  Upanishads:	  Breath	  of	  the	  eternal:	  The	  Teitan	  Press,	  Inc.	  Pruessmann	  KP,	  Weiger	  M,	  Scheidegger	  MB,	  Boesiger	  P	  (1999)	  SENSE:	  sensitivity	  encoding	  for	  fast	  MRI.	  Magnetic	  resonance	  in	  medicine	  42:952-­‐962.	  	   195	  Qin	  Y-­‐L,	  Mcnaughton	  BL,	  Skaggs	  WE,	  Barnes	  CA	  (1997)	  Memory	  reprocessing	  in	  corticocortical	  and	  hippocampocortical	  neuronal	  ensembles.	  Philosophical	  Transactions	  of	  the	  Royal	  Society	  B:	  Biological	  Sciences	  352:1525-­‐1533.	  Race	  E,	  Keane	  MM,	  Verfaellie	  M	  (2011)	  Medial	  temporal	  lobe	  damage	  causes	  deficits	  in	  episodic	  memory	  and	  episodic	  future	  thinking	  not	  attributable	  to	  deficits	  in	  narrative	  construction.	  The	  Journal	  of	  Neuroscience	  31:10262-­‐10269.	  Race	  E,	  Keane	  MM,	  Verfaellie	  M	  (2013)	  Losing	  sight	  of	  the	  future:	  Impaired	  semantic	  prospection	  following	  medial	  temporal	  lobe	  lesions.	  Hippocampus	  23:268-­‐277.	  Radua	  J,	  Mataix-­‐Cols	  D	  (2009)	  Voxel-­‐wise	  meta-­‐analysis	  of	  grey	  matter	  changes	  in	  obsessive–compulsive	  disorder.	  The	  British	  Journal	  of	  Psychiatry	  195:393-­‐402.	  Raichle	  ME	  (2010)	  Two	  views	  of	  brain	  function.	  Trends	  Cogn	  Sci	  14:180-­‐190.	  Raichle	  ME,	  MacLeod	  AM,	  Snyder	  AZ,	  Powers	  WJ,	  Gusnard	  DA,	  Shulman	  GL	  (2001)	  A	  default	  mode	  of	  brain	  function.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  USA	  98:678-­‐682.	  Raichle	  ME,	  Mintun	  MA	  (2006)	  Brain	  work	  and	  brain	  imaging.	  Annu	  Rev	  Neurosci	  29:449-­‐476.	  Ramón	  y	  Cajal	  S	  (1928)	  Degeneration	  &	  regeneration	  of	  the	  nervous	  system:	  Oxford	  University	  Press,	  Humphrey	  Milford.	  Rechtschaffen	  A,	  Kales	  A	  (1968)	  A	  manual	  of	  standardized	  terminology,	  techniques	  and	  scoring	  system	  for	  sleep	  stages	  of	  human	  subjects.	  Rolls	  ET	  (2000)	  Hippocampo-­‐cortical	  and	  cortico-­‐cortical	  backprojections.	  Hippocampus	  10:380-­‐388.	  	   196	  Rosenbaum	  RS,	  Gilboa	  A,	  Levine	  B,	  Winocur	  G,	  Moscovitch	  M	  (2009)	  Amnesia	  as	  an	  impairment	  of	  detail	  generation	  and	  binding:	  Evidence	  from	  personal,	  fictional,	  and	  semantic	  narratives	  in	  KC.	  Neuropsychologia	  47:2181-­‐2187.	  Rothbaum	  BO,	  Meadows	  EA,	  Resick	  P,	  Foy	  DW	  (2000)	  Cognitive-­‐behavioral	  therapy.	  Rubin	  RD,	  Watson	  PD,	  Duff	  MC,	  Cohen	  NJ	  (2014)	  The	  role	  of	  the	  hippocampus	  in	  flexible	  cognition	  and	  social	  behavior.	  Frontiers	  in	  Human	  Neuroscience	  8.	  Rueckert	  D,	  Sonoda	  LI,	  Hayes	  C,	  Hill	  DL,	  Leach	  MO,	  Hawkes	  DJ	  (1999)	  Nonrigid	  registration	  using	  free-­‐form	  deformations:	  application	  to	  breast	  MR	  images.	  Medical	  Imaging,	  IEEE	  Transactions	  on	  18:712-­‐721.	  Sagi	  Y,	  Tavor	  I,	  Hofstetter	  S,	  Tzur-­‐Moryosef	  S,	  Blumenfeld-­‐Katzir	  T,	  Assaf	  Y	  (2012)	  Learning	  in	  the	  fast	  lane:	  new	  insights	  into	  neuroplasticity.	  Neuron	  73:1195-­‐1203.	  Salimi-­‐Khorshidi	  G,	  Smith	  SM,	  Keltner	  JR,	  Wager	  TD,	  Nichols	  TE	  (2009)	  Meta-­‐analysis	  of	  neuroimaging	  data:	  a	  comparison	  of	  image-­‐based	  and	  coordinate-­‐based	  pooling	  of	  studies.	  NeuroImage	  45:810-­‐823.	  Sämann	  PG,	  Wehrle	  R,	  Hoehn	  D,	  Spoormaker	  VI,	  Peters	  H,	  Tully	  C,	  Holsboer	  F,	  Czisch	  M	  (2011)	  Development	  of	  the	  brain's	  default	  mode	  network	  from	  wakefulness	  to	  slow	  wave	  sleep.	  Cerebral	  Cortex	  bhq295.	  Schacter	  DL,	  Addis	  DR,	  Buckner	  RL	  (2007)	  Remembering	  the	  past	  to	  imagine	  the	  future:	  the	  prospective	  brain.	  Nature	  Reviews	  Neuroscience	  8:657-­‐661.	  Schacter	  DL,	  Addis	  DR,	  Hassabis	  D,	  Martin	  VC,	  Spreng	  RN,	  Szpunar	  KK	  (2012)	  The	  future	  of	  memory:	  remembering,	  imagining,	  and	  the	  brain.	  Neuron	  76:677-­‐694.	  Schindler	  R	  (1953)	  Das	  Traumleben	  der	  Leukotomierten.	  Zeitschr	  Nervenheil	  6:330-­‐334.	  	   197	  Schmahmann	  JD,	  Sherman	  JC	  (1998)	  The	  cerebellar	  cognitive	  affective	  syndrome.	  Brain	  121:561-­‐579.	  Schooler	  JW,	  Mrazek	  MD,	  Franklin	  MS,	  Baird	  B,	  Mooneyham	  BW,	  Zedelius	  C,	  Broadway	  JM	  (2014)	  The	  Middle	  Way:	  Finding	  the	  Balance	  between	  Mindfulness	  and	  Mind-­‐Wandering.	  Schooler	  JW,	  Smallwood	  J,	  Christoff	  K,	  Handy	  TC,	  Reichle	  ED,	  Sayette	  MA	  (2011)	  Meta-­‐awareness,	  perceptual	  decoupling	  and	  the	  wandering	  mind.	  Trends	  Cogn	  Sci	  15:319-­‐326.	  Schopenhauer	  A,	  Payne	  EFJ	  (1974)	  Parerga	  and	  Paralipomena:	  short	  philosophical	  essays:	  Oxford	  University	  Press.	  Schredl	  M	  (2000)	  Continuity	  between	  waking	  life	  and	  dreaming:	  Are	  all	  waking	  activities	  reflected	  equally	  often	  in	  dreams?	  Perceptual	  and	  motor	  skills	  90:844-­‐846.	  Schredl	  M,	  Erlacher	  D	  (2004)	  Lucid	  dreaming	  frequency	  and	  personality.	  Personality	  and	  Individual	  Differences	  37:1463-­‐1473.	  Schredl	  M,	  Erlacher	  D	  (2011)	  Frequency	  of	  lucid	  dreaming	  in	  a	  representative	  German	  sample.	  Perceptual	  and	  motor	  skills	  112:104-­‐108.	  Schulz	  R,	  Woermann	  FG,	  Ebner	  A	  (2007)	  When	  written	  words	  become	  moving	  pictures:	  complex	  visual	  hallucinations	  on	  stimulation	  of	  the	  lateral	  occipital	  lobe.	  Epilepsy	  &	  Behavior	  11:147-­‐151.	  Scoville	  WB,	  Milner	  B	  (1957)	  Loss	  of	  recent	  memory	  after	  bilateral	  hippocampal	  lesions.	  Journal	  of	  neurology,	  neurosurgery,	  and	  psychiatry	  20:11.	  	   198	  Selimbeyoglu	  A,	  Parvizi	  J	  (2010)	  Electrical	  stimulation	  of	  the	  human	  brain:	  perceptual	  and	  behavioral	  phenomena	  reported	  in	  the	  old	  and	  new	  literature.	  Frontiers	  in	  Human	  Neuroscience	  4.	  Sergio	  W	  (1988)	  Use	  of	  DMAE	  (2-­‐dimethylaminoethanol)	  in	  the	  induction	  of	  lucid	  dreams.	  Medical	  hypotheses	  26:255-­‐257.	  Sharma	  A	  (2012)	  Sleep	  as	  a	  State	  of	  Consciousness	  in	  Advaita	  Vedanta:	  SUNY	  Press.	  Sheline	  YI,	  Barch	  DM,	  Price	  JL,	  Rundle	  MM,	  Vaishnavi	  SN,	  Snyder	  AZ,	  Mintun	  MA,	  Wang	  S,	  Coalson	  RS,	  Raichle	  ME	  (2009)	  The	  default	  mode	  network	  and	  self-­‐referential	  processes	  in	  depression.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  106:1942-­‐1947.	  Shepperd	  J,	  Malone	  W,	  Sweeny	  K	  (2008)	  Exploring	  causes	  of	  the	  self‐serving	  bias.	  Social	  and	  Personality	  Psychology	  Compass	  2:895-­‐908.	  Shulman	  GL,	  Fiez	  JA,	  Corbetta	  M,	  Buckner	  RL,	  Miezin	  FM,	  Raichle	  ME,	  Petersen	  SE	  (1997)	  Common	  blood	  flow	  changes	  across	  visual	  tasks:	  II.	  Decreases	  in	  cerebral	  cortex.	  Journal	  of	  Cognitive	  Neuroscience	  9:648-­‐663.	  Silber	  MH,	  Ancoli-­‐Israel	  S,	  Bonnet	  MH,	  Chokroverty	  S,	  Grigg-­‐Damberger	  MM,	  Hirshkowitz	  M,	  Kapen	  S,	  Keenan	  SA,	  Kryger	  MH,	  Penzel	  T	  (2007)	  The	  visual	  scoring	  of	  sleep	  in	  adults.	  J	  Clin	  Sleep	  Med	  3:121-­‐131.	  Silvia	  PJ,	  Beaty	  RE,	  Nusbaum	  EC,	  Eddington	  KM,	  Levin-­‐Aspenson	  H,	  Kwapil	  TR	  (2014)	  Everyday	  creativity	  in	  daily	  life:	  An	  experience-­‐sampling	  study	  of	  “little	  c”	  creativity.	  Psychology	  of	  Aesthetics,	  Creativity,	  and	  the	  Arts	  8:183.	  Simons	  J,	  Spiers	  H	  (2003)	  Prefrontal	  and	  medial	  temporal	  lobe	  interactions	  in	  long-­‐term	  memory.	  Nature	  Reviews	  Neuroscience	  4:637-­‐648.	  	   199	  Simonton	  DK	  (1999a)	  Creativity	  as	  blind	  variation	  and	  selective	  retention:	  Is	  the	  creative	  process	  Darwinian?	  Psychological	  Inquiry	  10:309-­‐328.	  Simonton	  DK	  (1999b)	  Origins	  of	  genius:	  Darwinian	  perspectives	  on	  creativity:	  Oxford	  University	  Press.	  Singer	  JL	  (1961)	  Imagination	  and	  waiting	  ability	  in	  young	  children.	  Journal	  of	  personality	  29:396-­‐413.	  Singer	  JL	  (1966)	  Daydreaming:	  An	  introduction	  to	  the	  experimental	  study	  of	  inner	  experience.	  New	  York:	  Crown	  Publishing	  Group/Random	  House.	  Singer	  JL,	  Antrobus	  JS	  (1970)	  Imaginal	  Process	  Inventory.	  Singer	  JL,	  Antrobus	  JS	  (1972)	  Daydreaming,	  imaginal	  processes,	  and	  personality:	  A	  normative	  study.	  The	  function	  and	  nature	  of	  imagery	  175-­‐202.	  Singer	  JL,	  Brown	  S	  (1960)	  The	  experience	  type:	  Some	  behavioral	  correlates	  and	  theoretical	  implications.	  Rorschach	  psychology	  New	  York:	  Wiley	  223-­‐259.	  Singer	  JL,	  McCraven	  V	  (1961)	  Some	  characteristics	  of	  adult	  daydreaming.	  The	  Journal	  of	  Psychology	  51:151-­‐164.	  Singer	  JL,	  McCraven	  VG	  (1962)	  Daydreaming	  patterns	  of	  American	  subcultural	  groups.	  International	  Journal	  of	  Social	  Psychiatry	  8:272-­‐282.	  Smallwood	  J,	  Gorgolewski	  KJ,	  Golchert	  J,	  Ruby	  FJ,	  Engen	  H,	  Baird	  B,	  Vinski	  MT,	  Schooler	  JW,	  Margulies	  DS	  (2013a)	  The	  default	  modes	  of	  reading:	  modulation	  of	  posterior	  cingulate	  and	  medial	  prefrontal	  cortex	  connectivity	  associated	  with	  comprehension	  and	  task	  focus	  while	  reading.	  Frontiers	  in	  Human	  Neuroscience	  7.	  	   200	  Smallwood	  J,	  Nind	  L,	  O'Connor	  RC	  (2009)	  When	  is	  your	  head	  at?	  An	  exploration	  of	  the	  factors	  associated	  with	  the	  temporal	  focus	  of	  the	  wandering	  mind.	  Consciousness	  and	  cognition	  18:118-­‐125.	  Smallwood	  J,	  O'Connor	  RC	  (2011)	  Imprisoned	  by	  the	  past:	  unhappy	  moods	  lead	  to	  a	  retrospective	  bias	  to	  mind	  wandering.	  Cogn	  Emot	  25:1481-­‐1490.	  Smallwood	  J,	  Schooler	  JW	  (2014)	  The	  Science	  of	  Mind	  Wandering:	  Empirically	  Navigating	  the	  Stream	  of	  Consciousness.	  Annual	  review	  of	  psychology	  66.	  Smallwood	  J,	  Tipper	  C,	  Brown	  K,	  Baird	  B,	  Engen	  H,	  Michaels	  JR,	  Grafton	  S,	  Schooler	  JW	  (2013b)	  Escaping	  the	  here	  and	  now:	  evidence	  for	  a	  role	  of	  the	  default	  mode	  network	  in	  perceptually	  decoupled	  thought.	  NeuroImage	  69:120-­‐125.	  Smith	  SM	  (2002)	  Fast	  robust	  automated	  brain	  extraction.	  Human	  Brain	  Mapping	  17:143-­‐155.	  Smith	  SM,	  Jenkinson	  M,	  Woolrich	  MW,	  Beckmann	  CF,	  Behrens	  TE,	  Johansen-­‐Berg	  H,	  Bannister	  PR,	  De	  Luca	  M,	  Drobnjak	  I,	  Flitney	  DE	  (2004)	  Advances	  in	  functional	  and	  structural	  MR	  image	  analysis	  and	  implementation	  as	  FSL.	  NeuroImage	  23:S208-­‐S219.	  Snyder	  TJ,	  Gackenbach	  J	  (1988)	  Individual	  differences	  associated	  with	  lucid	  dreaming.	  In:	  Conscious	  mind,	  sleeping	  brain,	  pp	  221-­‐259:	  Springer.	  Solms	  M	  (1997)	  The	  neuropsychology	  of	  dreams:	  A	  clinico-­‐anatomical	  study:	  Lawrence	  Erlbaum	  Associates	  Publishers.	  Solms	  M	  (2000a)	  Dreaming	  and	  REM	  sleep	  are	  controlled	  by	  different	  brain	  mechanisms.	  Behavioral	  and	  Brain	  Sciences	  23:843-­‐850.	  	   201	  Solms	  M	  (2000b)	  Dreaming	  and	  REM	  sleep	  are	  controlled	  by	  different	  brain	  mechanisms.	  Behavioral	  and	  Brain	  Sciences	  23:843-­‐850.	  Song	  X,	  Wang	  X	  (2012)	  Mind	  wandering	  in	  Chinese	  daily	  lives–an	  experience	  sampling	  study.	  PLoS	  ONE	  7:e44423.	  Spiers	  H,	  Maguire	  E	  (2006a)	  Spontaneous	  mentalizing	  during	  an	  interactive	  real	  world	  task:	  An	  fMRI	  study.	  Neuropsychologia	  44:1674-­‐1682.	  Spiers	  H,	  Maguire	  E	  (2006b)	  Thoughts,	  behaviour,	  and	  brain	  dynamics	  during	  navigation	  in	  the	  real	  world.	  NeuroImage	  31:1826-­‐1840.	  Spreng	  RN,	  Mar	  RA,	  Kim	  AS	  (2009)	  The	  common	  neural	  basis	  of	  autobiographical	  memory,	  prospection,	  navigation,	  theory	  of	  mind,	  and	  the	  default	  mode:	  a	  quantitative	  meta-­‐analysis.	  Journal	  of	  Cognitive	  Neuroscience	  21:489-­‐510.	  Spreng	  RN,	  Sepulcre	  J,	  Turner	  GR,	  Stevens	  WD,	  Schacter	  DL	  (2013)	  Intrinsic	  architecture	  underlying	  the	  relations	  among	  the	  default,	  dorsal	  attention,	  and	  frontoparietal	  control	  networks	  of	  the	  human	  brain.	  Journal	  of	  Cognitive	  Neuroscience	  25:74-­‐86.	  Spreng	  RN,	  Stevens	  WD,	  Chamberlain	  JP,	  Gilmore	  AW,	  Schacter	  DL	  (2010)	  Default	  network	  activity,	  coupled	  with	  the	  frontoparietal	  control	  network,	  supports	  goal-­‐directed	  cognition.	  NeuroImage	  53:303-­‐317.	  Squire	  LR	  (2004)	  Memory	  systems	  of	  the	  brain:	  a	  brief	  history	  and	  current	  perspective.	  Neurobiology	  of	  learning	  and	  memory	  82:171-­‐177.	  Stawarczyk	  D,	  Cassol	  H,	  D'Argembeau	  A	  (2013)	  Phenomenology	  of	  future-­‐oriented	  mind-­‐wandering	  episodes.	  Frontiers	  in	  Psychology	  4.	  	   202	  Stawarczyk	  D,	  D'Argembeau	  A	  (2015)	  Neural	  correlates	  of	  personal	  goal	  processing	  during	  episodic	  future	  thinking	  and	  mind‐wandering:	  An	  ALE	  meta‐analysis.	  Human	  Brain	  Mapping	  36:2928-­‐2947.	  Stawarczyk	  D,	  Majerus	  S,	  Maj	  M,	  Van	  der	  Linden	  M,	  D'Argembeau	  A	  (2011a)	  Mind-­‐wandering:	  phenomenology	  and	  function	  as	  assessed	  with	  a	  novel	  experience	  sampling	  method.	  Acta	  Psychol	  (Amst)	  136:370-­‐381.	  Stawarczyk	  D,	  Majerus	  S,	  Maquet	  P,	  D'Argembeau	  A	  (2011b)	  Neural	  correlates	  of	  ongoing	  conscious	  experience:	  both	  task-­‐unrelatedness	  and	  stimulus-­‐independence	  are	  related	  to	  default	  network	  activity.	  PLoS	  One	  6:e16997.	  Steele	  JD,	  Christmas	  D,	  Eljamel	  MS,	  Matthews	  K	  (2008)	  Anterior	  cingulotomy	  for	  major	  depression:	  clinical	  outcome	  and	  relationship	  to	  lesion	  characteristics.	  Biological	  psychiatry	  63:670-­‐677.	  Stenberg	  D	  (2007)	  Neuroanatomy	  and	  neurochemistry	  of	  sleep.	  Cellular	  and	  Molecular	  Life	  Sciences	  64:1187-­‐1204.	  Stenstrom	  P,	  Fox	  KCR,	  Solomonova	  E,	  Nielsen	  TA	  (2012)	  Mentation	  during	  sleep	  onset	  theat	  bursts	  in	  a	  trained	  participant:	  A	  role	  for	  NREM	  stage	  1	  sleep	  in	  memory	  processing?	  International	  Journal	  of	  Dream	  Research	  5:37-­‐46.	  Stoodley	  CJ,	  Schmahmann	  JD	  (2009)	  Functional	  topography	  in	  the	  human	  cerebellum:	  a	  meta-­‐analysis	  of	  neuroimaging	  studies.	  NeuroImage	  44:489-­‐501.	  Striedter	  GF	  (2005)	  Principles	  of	  brain	  evolution:	  Sinauer	  Associates.	  Stumbrys	  T,	  Erlacher	  D,	  Schädlich	  M,	  Schredl	  M	  (2012)	  Induction	  of	  lucid	  dreams:	  A	  systematic	  review	  of	  evidence.	  Consciousness	  and	  cognition	  21:1456-­‐1475.	  	   203	  Stumbrys	  T,	  Erlacher	  D,	  Schredl	  M	  (2013)	  Testing	  the	  involvement	  of	  the	  prefrontal	  cortex	  in	  lucid	  dreaming:	  a	  tDCS	  study.	  Consciousness	  and	  cognition	  22:1214-­‐1222.	  Svoboda	  E,	  McKinnon	  MC,	  Levine	  B	  (2006)	  The	  functional	  neuroanatomy	  of	  autobiographical	  memory:	  a	  meta-­‐analysis.	  Neuropsychologia	  44:2189-­‐2208.	  Szpunar	  KK,	  Spreng	  RN,	  Schacter	  DL	  (2014)	  A	  taxonomy	  of	  prospection:	  Introducing	  an	  organizational	  framework	  for	  future-­‐oriented	  cognition.	  Proceedings	  of	  the	  National	  Academy	  of	  Sciences	  111:18414-­‐18421.	  Tanaka	  H,	  Hayashi	  M,	  Hori	  T	  (1996)	  Statistical	  features	  of	  hypnagogic	  EEG	  measured	  by	  a	  new	  scoring	  system.	  Sleep	  19:731-­‐738.	  Tanaka	  H,	  Hayashi	  M,	  Hori	  T	  (1997)	  Topographical	  characteristics	  and	  principal	  component	  structure	  of	  the	  hypnagogic	  EEG.	  Sleep	  20:523-­‐534.	  Teyler	  TJ,	  DiScenna	  P	  (1985)	  The	  role	  of	  hippocampus	  in	  memory:	  a	  hypothesis.	  Neuroscience	  &	  Biobehavioral	  Reviews	  9:377-­‐389.	  Teyler	  TJ,	  DiScenna	  P	  (1986)	  The	  hippocampal	  memory	  indexing	  theory.	  Behavioral	  neuroscience	  100:147.	  Teyler	  TJ,	  Rudy	  JW	  (2007)	  The	  hippocampal	  indexing	  theory	  and	  episodic	  memory:	  updating	  the	  index.	  Hippocampus	  17:1158-­‐1169.	  Thach	  W	  (1996)	  On	  the	  specific	  role	  of	  the	  cerebellum	  in	  motor	  learning	  and	  cognition:	  Clues	  from	  PET	  activation	  and	  lesion	  studies	  in	  man.	  Behavioral	  and	  Brain	  Sciences	  19:411-­‐433.	  Thompson	  E	  (2014)	  Waking,	  Dreaming,	  Being:	  Self	  and	  Consciousness	  in	  Neuroscience,	  Meditation,	  and	  Philosophy:	  Columbia	  University	  Press.	  	   204	  Torda	  C	  (1969)	  Dreams	  of	  subjects	  with	  bilateral	  hippocampal	  lesions.	  Acta	  Psychiatrica	  Scandinavica	  45:277-­‐288.	  Tuch	  DS,	  Reese	  TG,	  Wiegell	  MR,	  Makris	  N,	  Belliveau	  JW,	  Wedeen	  VJ	  (2002)	  High	  angular	  resolution	  diffusion	  imaging	  reveals	  intravoxel	  white	  matter	  fiber	  heterogeneity.	  Magnetic	  resonance	  in	  medicine	  48:577-­‐582.	  Turkeltaub	  PE,	  Eden	  GF,	  Jones	  KM,	  Zeffiro	  TA	  (2002)	  Meta-­‐analysis	  of	  the	  functional	  neuroanatomy	  of	  single-­‐word	  reading:	  method	  and	  validation.	  NeuroImage	  16:765-­‐780.	  Turkeltaub	  PE,	  Eickhoff	  SB,	  Laird	  AR,	  Fox	  M,	  Wiener	  M,	  Fox	  P	  (2012)	  Minimizing	  within‐experiment	  and	  within‐group	  effects	  in	  activation	  likelihood	  estimation	  meta‐analyses.	  Human	  Brain	  Mapping	  33:1-­‐13.	  Tusche	  A,	  Smallwood	  J,	  Bernhardt	  BC,	  Singer	  T	  (2014)	  Classifying	  the	  wandering	  mind:	  Revealing	  the	  affective	  content	  of	  thoughts	  during	  task-­‐free	  rest	  periods.	  NeuroImage	  97:107-­‐116.	  Vanhaudenhuyse	  A,	  Demertzi	  A,	  Schabus	  M,	  Noirhomme	  Q,	  Bredart	  S,	  Boly	  M,	  Phillips	  C,	  Soddu	  A,	  Luxen	  A,	  Moonen	  G,	  Laureys	  S	  (2011)	  Two	  distinct	  neuronal	  networks	  mediate	  the	  awareness	  of	  environment	  and	  of	  self.	  Journal	  of	  Cognitive	  Neuroscience	  23:570-­‐578.	  Vignal	  J-­‐P,	  Maillard	  L,	  McGonigal	  A,	  Chauvel	  P	  (2007)	  The	  dreamy	  state:	  hallucinations	  of	  autobiographic	  memory	  evoked	  by	  temporal	  lobe	  stimulations	  and	  seizures.	  Brain	  130:88-­‐99.	  	   205	  Vincent	  JL,	  Kahn	  I,	  Snyder	  AZ,	  Raichle	  ME,	  Buckner	  RL	  (2008)	  Evidence	  for	  a	  frontoparietal	  control	  system	  revealed	  by	  intrinsic	  functional	  connectivity.	  Journal	  of	  neurophysiology	  100:3328-­‐3342.	  Vogel	  G,	  Foulkes	  D,	  Trosman	  H	  (1966)	  Ego	  functions	  and	  dreaming	  during	  sleep	  onset.	  Archives	  of	  General	  Psychiatry	  14:238-­‐248.	  Vogt	  BA,	  Berger	  GR,	  Derbyshire	  SW	  (2003)	  Structural	  and	  functional	  dichotomy	  of	  human	  midcingulate	  cortex.	  European	  Journal	  of	  Neuroscience	  18:3134-­‐3144.	  Voss	  U,	  Holzmann	  R,	  Hobson	  A,	  Paulus	  W,	  Koppehele-­‐Gossel	  J,	  Klimke	  A,	  Nitsche	  MA	  (2014)	  Induction	  of	  self	  awareness	  in	  dreams	  through	  frontal	  low	  current	  stimulation	  of	  gamma	  activity.	  Nature	  Neuroscience	  17:810-­‐812.	  Voss	  U,	  Holzmann	  R,	  Tuin	  I,	  Hobson	  JA	  (2009)	  Lucid	  dreaming:	  A	  state	  of	  consciousness	  with	  features	  of	  both	  waking	  and	  non-­‐lucid	  dreaming.	  Sleep	  32:1191-­‐1200.	  Wager	  TD,	  Lindquist	  M,	  Kaplan	  L	  (2007)	  Meta-­‐analysis	  of	  functional	  neuroimaging	  data:	  current	  and	  future	  directions.	  Social	  cognitive	  and	  affective	  neuroscience	  2:150-­‐158.	  Wager	  TD,	  Lindquist	  MA,	  Nichols	  TE,	  Kober	  H,	  Van	  Snellenberg	  JX	  (2009)	  Evaluating	  the	  consistency	  and	  specificity	  of	  neuroimaging	  data	  using	  meta-­‐analysis.	  NeuroImage	  45:S210-­‐221.	  Wakana	  S,	  Caprihan	  A,	  Panzenboeck	  MM,	  Fallon	  JH,	  Perry	  M,	  Gollub	  RL,	  Hua	  K,	  Zhang	  J,	  Jiang	  H,	  Dubey	  P	  (2007)	  Reproducibility	  of	  quantitative	  tractography	  methods	  applied	  to	  cerebral	  white	  matter.	  NeuroImage	  36:630-­‐644.	  Wang	  K,	  Yu	  C,	  Xu	  L,	  Qin	  W,	  Li	  K,	  Xu	  L,	  Jiang	  T	  (2009)	  Offline	  memory	  reprocessing:	  Involvement	  of	  the	  brain's	  default	  network	  in	  spontaneous	  thought	  processes.	  PLoS	  ONE	  4:e4867.	  	   206	  Wangyal	  T	  (1998)	  The	  Tibetan	  yogas	  of	  dream	  and	  sleep.	  Boston:	  Snow	  Lion	  Publications.	  Whitfield-­‐Gabrieli	  S,	  Ford	  JM	  (2012)	  Default	  mode	  network	  activity	  and	  connectivity	  in	  psychopathology.	  Annual	  review	  of	  clinical	  psychology	  8:49-­‐76.	  Wieser	  H	  (1979)	  Ictal	  “psychical	  phenomena”	  and	  stereoelectroencephalographic	  findings.	  In:	  EEG	  and	  clinical	  neurophysiology:	  proceedings	  of	  the	  second	  European	  congress	  on	  EEG	  and	  clinical	  neurophysiology,	  Salzburg,	  Austria	  Excerpta	  Medica,	  Amsterdam,	  pp	  62-­‐76.	  Williams	  RL,	  Karacan	  I,	  Hursch	  CJ	  (1974)	  Electroencephalography	  (EEG)	  of	  human	  sleep:	  clinical	  applications:	  John	  Wiley	  &	  Sons.	  Wilson	  MA,	  McNaughton	  BL	  (1994)	  Reactivation	  of	  hippocampal	  ensemble	  memories	  during	  sleep.	  Science	  265:676-­‐679.	  Wilson	  RS,	  Mayhew	  SD,	  Rollings	  DT,	  Goldstone	  A,	  Przezdzik	  I,	  Arvanitis	  TN,	  Bagshaw	  AP	  (2015)	  Influence	  of	  epoch	  length	  on	  measurement	  of	  dynamic	  functional	  connectivity	  in	  wakefulness	  and	  behavioural	  validation	  in	  sleep.	  NeuroImage	  112:169-­‐179.	  Wilson	  TD,	  Reinhard	  DA,	  Westgate	  EC,	  Gilbert	  DT,	  Ellerbeck	  N,	  Hahn	  C,	  Brown	  CL,	  Shaked	  A	  (2014)	  Just	  think:	  The	  challenges	  of	  the	  disengaged	  mind.	  Science	  345:75-­‐77.	  Windt	  JM	  (2010)	  The	  immersive	  spatiotemporal	  hallucination	  model	  of	  dreaming.	  Phenomenology	  and	  the	  Cognitive	  Sciences	  9:295-­‐316.	  Windt	  JM	  (2015)	  Dreaming:	  A	  conceptual	  framework	  for	  philosophy	  of	  mind	  and	  empirical	  research:	  MIT	  Press.	  	   207	  Yarkoni	  T	  (2009)	  Big	  Correlations	  in	  Little	  Studies:	  Inflated	  fMRI	  Correlations	  Reflect	  Low	  Statistical	  Power-­‐Commentary	  on	  Vul	  et	  al.	  (2009).	  Perspectives	  on	  Psychological	  Science	  4:294-­‐298.	  Yeo	  BTT,	  Kirienen	  FM,	  Sepulcre	  J,	  Sabuncu	  MR,	  Lashkari	  D,	  Hollinshead	  M,	  Roffman	  JL,	  Smoller	  JW,	  Zöllei	  L,	  Polimeni	  JR,	  Fischl	  B,	  Liu	  H,	  Buckner	  RL	  (2011)	  The	  organization	  of	  the	  human	  cerebral	  cortex	  estimated	  by	  intrinsic	  functional	  connectivity.	  Journal	  of	  neurophysiology	  106:1125-­‐1165.	  Yuschak	  T	  (2006)	  Advanced	  lucid	  dreaming.	  The	  power	  of	  supplements.	  Zadra	  AL,	  Donderi	  D,	  Pihl	  RO	  (1992)	  Efficacy	  of	  lucid	  dream	  induction	  for	  lucid	  and	  non-­‐lucid	  dreamers.	  Dreaming	  2:85.	  Zatorre	  RJ,	  Fields	  RD,	  Johansen-­‐Berg	  H	  (2012)	  Plasticity	  in	  gray	  and	  white:	  neuroimaging	  changes	  in	  brain	  structure	  during	  learning.	  Nat	  Neurosci	  15:528-­‐536.	  Zhiyan	  T,	  Singer	  JL	  (1997)	  Daydreaming	  styles,	  emotionality	  and	  the	  big	  five	  personality	  dimensions.	  Imagination,	  Cognition	  and	  Personality	  16:399-­‐414.	  	   

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0308694/manifest

Comment

Related Items