Endogenous and exogenous hemodynamic signals in primary visual cortex of alert non-human primates

Ma iana M. B. Ca doso Disse a ion p esen ed o ob ain he Ph.D deg ee in Biology | Neu oscience Ins i u o de Tecnologia Química e Biológica An ónio Xa ie | Uni e sidade No a de Lisboa Oei as, May, 2016 Inse he e an image wi h ounded co ne s Endogenous and exogenous hemodynamic signals in p ima y isual co ex o ale non-human p ima es Ma iana Ma celino Belchio Ca doso Disse a ion p esen ed o ob ain he Ph.D deg ee in Biology | Neu oscience Ins i u o de Tecnologia Química e Biológica An ónio Xa ie | Uni e sidade No a de Lisboa Oei as, May, 2016 Endogenous and exogenous hemodynamic signals in p ima y isual co ex o ale non-human p ima es Resea ch wo k coo dina ed by: ii Financial Suppo Financial suppo o his hesis was p o ided by Fundação pa a a Ciência e Tecnologia and Fundo Social Eu opeu h ough he Quad o Comuni á io de Apoio, doc o al ellowship: SFRH/BD/33276/2007. iii Acknowledgemen s Thank you o he supe -nine PGCN/INDP-2007: Í is, Isabel, José, Ma ga ida, Ma ia, Pa ícia, Pa ício, Ped o, Rod igo: colleagues and iends, o e he pas se e al yea s (be e no o name how many) we e o imp essi e suppo in di e en o ms and momen s. Thank you in pa icula o Ma ga ida wi hou whom he “chap e hesis” would ha e been ha d o imagine possible, and cleaning my ea s o e my poo w i ing. Thank you o Í is o he uncondi ional posi i ism. Thank you, Zé, o le ing me lea n o app ecia e you wonde ul imagina ion. Thank you o my ad iso , Ani uddha Das. Who kep his doo open a all imes. Tha hos ed me in his labo a o y o a e y, e y long ime; and was always pa ien wi h my slow p og ess. Thank you o he oppo uni y o wo king in you labo a o y and access o such ich se ups and da a. Thank you especially o B uss Lima and Ma ia (Masha) Bezlepkina. B uss was he mos pa ien eache I could ha e wished o . Wi h you, B uss, I go o lea n abou elec ophysiology and I wish I could say I had no lea n he sound o dying neu ons. Masha kep me sane when ha was a om a gi en, hank you o ha and he iendship. Thank you o he o he mem- be s in he lab, Ye geniy, I ha e lea ned a lo by o e lapping wi h you in he lab. Elena as well as Yana we e wonde ul ‘mo he s’ when I a i ed, hank you. Thank you as well o he wonde ul e e ina y eam, in pa icula hank you o D . Gi ma As aw, D . Rodol o Rica and D . Amy Cassano. In he hesis will no be included ea ly wo k done wi h oden s, bu he e we e a ew c i ical inspi ing people ha should s ill be acknowledged he e: Ma k Ande mann, Daniel W. Wesson, Tomas H omadka, Dinu Flo in Albeanu, Ashesh K. Dhawale. Also Ma c Bucklin om Ani uddha Das’s lab, ha was always eady o a challenge. Thank you o wonde ul Do o hée, and he awesome husband A naud, whose ime and good hea ing skills helped me keep ocused on wha my i p io i ies migh be. Thank you, Angelie, wi h whom all is possible. Thank you F anco, i I had an old b o he , I wish he would ha e been like you. Thank you o he ea ly iends wi h whom I lea ned abou he Big ci y: Gil, Joe, Yan and Rena a: he i s hos . Thank o he iends ha keep checking on me om a a , Ma alda, Fá ima, Inês, Mónica, Ped o, Joana, Miguel, Lena. Thank you, Zach, Rui, Ma a, o c ea ing his wonde ul p og am and scien i ic en i onmen and gi ing me he oppo uni y o be pa o i . Thank you o my hesis commi ee, Ch is ian and Michael, who we e always suppo i e, posi i e and encou aging. Thank you o he hos ing ins i u ion, Columbia Uni e si y, and speci ically o a ew people ha handled my p ocess e icien ly and wi h a smile, F ed Lowe and Alla Ke zhne . Thank you also o he Champalimaud s uc u e and o ganiza ion: o he g ea lexibili y and ease o wo king wi h. Thank you, S ephan, o suppo and pa ience h ough he leas un imes. Thank you o my amily, Adelino, Isabel, Ma a and Samuel, o he uncondi- ional suppo (e en inancial), help, ene gy and us ! Finally, hank you o he animals: no on he class o human p ima es, and no olun ee on his p ocess, bu wi hou whom his p ojec would ha e no been possible. Resumo O uso de écnicas de imagiologia ce eb al (neu oimagem) em pa icula aquelas adequadas ao uso em humanos em expe ienciado ápida dissemi- nação. E em ha ido g ande c escimen o nas á eas de in es igação que es udam a co elação en e ac i idade neu onal e p ocessos cogni i os. No en an o os undamen os neu onais dos sinais esul an es de neu oimagem não es ão ainda o almen e ca ac e izados; es a é ainda uma á ea de in ensa in es igação. E no con ex o de a e as compo amen ais/cogni i as, a in e p e ação de neu oimagem é ainda mais complexa. Em sis emas senso iais p imá ios, há a pe spec i a de que as espos as de neu oimagem e lec em sob e udo in o mação sob e es ímulos senso iais ex e nos. Há as a in es igação que se oca no mapeamen o de es ímulos e classes de es ímulos no có ex. No en an o, mesmo sis emas senso iais p imá ios podem con e in o mação sob e compo amen o, como po exemplo es ádios de compo amen o. Po exemplo, embo a o có ex isual p imá io esponda p incipalmen e a es ímulos isuais, ambém é modulado po espos as como a enção. A o ma como es as espos as es ão codi ica- das é undamen al pa a a in e p e ação de neu oimagem. Igualmen e, comp eende como di e en es aspec os das espos as de neu oimagem es ão elacionados com a ac i idade neu onal subjacen e é c ucial pa a a in e p e ação de neu oimagem. No labo a ó io inha sido iden i icada uma espos a hemodinâmica elacio- nada com execução de a e as ( ask- ela ed), no có ex isual p imá io de p ima as execu ando uma a e a de ixação isual pe iódica. Es e sinal oi obse ado na ausência de es imulação isual di ec a (não ha endo po an o es ímulo isual ou espos a isual a es ímulos). Não se obse a am al e ações na quan idade de po enciais de acção dos neu ónios na egião de onde se epo ou a ac i idade hemodinâmica. No en an o, nes as a e as a espos a hemodinâmica elacionada com a a e a é obus a e es a i espos a ajus a-se à du ação da a e a. Es a espos a con i ma a p esença de uma espos a endógena numa á ea senso ial p imá ia. A ausência de co elação com a ac i i ade neu onal (medida como al e ações na equência de po enciais de acção) ein o ça a impo ância de comp eende os mecanismos que es ão na base da espos a hemodinâmica. O abalho ap esen ado nes a ese em como objec i o auxilia a in e p e a- ção de neu oimagem, em pa icula en a dis ingui con ibuições endóge- nas e exógenas. Seguimos dois caminhos expe imen ais: um que en a ca ac e iza as espos as hemodinâmicas elacionas com a ap esen ação de es ímulos, e ou a cujo objec i o é de comp eende con ibuições endógenas pa a o sinal. Em ambas as ci cuns âncias, pa a além de neu oimagem, oi medida ac i idade neu onal (na o ma de po enciais de acção locais, LFP, ou equência de po enciais de acção neu onais locais). G a ámos simul aneamen e ac i idade de neu oimagem bem como de ele o isiologia em p ima as a execu a a e as pe iódicas. A écnica de neu oimagem u ilizada, imagiologia óp ica de sinais in ínsecos, baseia-se na abso ção p e e encial de luz isí el pela hemoglobina p esen e em ecidos; a a-se po an o uma medida indi ec a de me abolismo. Recolhemos dados no có ex isual p imá io, uma egião co ical bem conhecida, com espos as neu onais a es ímulos bem ca ac e izada e que ana omicamen e se localiza na supe ície do cé eb o, po an o ajus ada ao uso de imagem óp ica in ínseca. Dada a mon agem expe imen al disponí el no labo a ó io, há a possibili- dade de compa a dados de neu oimagem e elec o isiologia adqui idos simul aneaman e (não ípico aquando o uso de ou as écnicas de neu oi- magem) o que nos coloca numa posição p i eligiada pa a a alia a elação en e neu oimagem e ac i idade neu onal. U ilizámos uma a e a pe iódica com a ap esen ação de es ímulos isuais com uma elação bem es abelecida en e p op iedades do es ímulo e ac i idade neu onal: usámos con as e do es ímulo pa a es a uma gama de in ensidades de ac i idade neu onal. Con as e e ac i idade elec o isiológica êm uma elação mono ó- ii nica: aumen os no con as e es ão associados a aumen os na equência de po enciais de acção (obedecendo uma unção hipe bólica). Ao u ilizá mos con as e pa a induzi al e ações p e isí eis em ac i idade neu onal, obse ámos que numa a e a pe iódica a espos a de neu oimagem e lec e a soma linea de uma componen e exógena ( elacionada com o es ímulo) e ou a endógena ( elacionada com a a e a). A componen e elacionada com o es ímulo em uma elação linea com a ac i idade neu onal local (medida como al e ações na equência de po enciais de acção). Finalmen e, que iamos ambém a alia po enciais con ibuições pa a a ac o es da espos a endógena (independen e do es ímulo) que possam con ibui pa a o sinal de neu oimagem. Usámos a mesma a e a já an es u ilizada do labo a ó io: ixação na ausência de es imulação isual (pa a além do pon o de ixação), obse ámos pe íodos em que o animal não es a a en ol ido na a e a es a am associados com al e ações len as na espos a hemodinâmica. Pa a além da já mencionada espos a hemodinâ- mica elacionada com execução de a e as ( ask- ela ed), p opomos que o en ol imen o (engagemen ) numa a e a em uma con ibuição signi ica i a pa a al e ações len as no nosso sinal de neu oimagem. Obse ámos aumen os na espos a hemodinâmica acompanhados de diminuição do ba imen o ca díaco e ligei a diminuição na equência dos po enciais de acção, no có ex isual p imá io. As al e ações na espos a hemodinâmica são signi ica i as em magni ude. O abalho ap esen ado nes a ese a ança assim a nossa comp eensão da na u eza da neu oimagem. iii Abs ac The ad en o neu oimaging echniques in pa icula he ones sui able o s udies in ale humans has dissemina ed as . Resea ch in ields in ol ing neu o-co ela es o cogni i e p ocesses has lou ished. S ill he neu al unde pinnings o he neu oimaging signals emain o be ully cha ac e ized; his ield is an ac i e opic o esea ch. In he con ex o beha io /cogni ion, he in e p e a ion o neu oimaging signals is e en mo e in ica e. In ea ly senso y sys ems, neu oimaging signals a e hough o p ima ily ca - y in o ma ion abou senso y inpu s; he e is signi ican esea ch ocused on mapping e e mo e speci ic s imuli and s imulus classes o co ical egions. Ne e heless e en ea ly senso y sys ems can elay in o ma ion on beha io , like b ain s a es. Fo example, i is known ha isual co ex e en hough p ima ily esponding o isual s imuli, is also sensi i e o such signals as a en ion. The way hese signals a e di e en ially encoded is c i ical in ying o in e p e neu oimaging signals. Also how he di e en aspec s o neu- oimaging signals ela ed o unde lying neu onal ac i i y is c i ical o neu oimaging in e p e a ion. Ea lie he lab showed he exis ence o a hemodynamic ask- ela ed signal, in p ima y isual co ex o ale non-human p ima es pe o ming a pe iodic ixa ion ask. This signal was obse ed in he absence o di ec isual s imula ion ( he e was no s imulus p esen o a s imulus esponse). In he icini y o he eco ded hemodynamic esponse, he e we e no changes in he spiking a es o neu ons, which could p edic he homodynamic ask- ela ed signal. Ne e heless he e was a obus ask- ela ed hemodynamic esponse, which en ains o ial schedule. I con i ms he p esence o an endogenous esponse in an ea ly senso y a ea. The lack o a local neu onal co ela e (as changes in i ing a e) u he emphasizes he impo ance o unde s anding he mechanisms unde lying he hemodynamic signals. x 2.3.10 C oss- alida ion o HRFSTIM ke nels ac oss sessions .............. 20 2.3.11 Decon olu ion ............................................................................. 20 2.3.12 Checking he s abili y o ou p ima y indings agains a iabili y in elec ode eco dings .............................................................................. 21 2.4 Resul s................................................................................................. 22 2.4.1 Spikes poo ly p edic hemodynamics in pe iodic ask ............. 23 2.4.2 Modi ied linea model wi h wo signal componen s.................. 25 2.4.3 T ial- ela ed signal consis en in s imulus and da k oom ....... 31 2.4.4 Spikes poo ly p edic blank- ial and da k- oom signals .......... 36 2.4.5 Blank sub ac ion equi ed o es ima e spikes om imaging... 40 2.5 Discussion ........................................................................................... 44 2.6 Supplemen a y in o ma ion ................................................................ 48 2.6.1 Appendix: Homogeneous Linea (‘Null’) and Modi ied Linea Model (MLM) o Spike-p edic ed Hemodynamics Appendix: Homogeneous Linea (‘Null’) and Modi ied Linea Model (MLM) o Spike-p edic ed Hemodynamics ............................................................... 48 2.6.2 Supplemen a y Figu es .............................................................. 52 2.7 Acknowledgemen s ............................................................................ 59 3 The hemodynamic ask- ela ed signal in p ima y isual co ex o ale non-human p ima es pe o ming simple asks ................................................ 61 3.1 In oduc ion ......................................................................................... 61 3.2 Me hods ............................................................................................... 63 3.3 Va iable T ial Du a ions ..................................................................... 64 3.4 Va iable Fixa ion Du a ions ............................................................... 73 3.5 Va iable Rewa d Amoun ................................................................... 79 3.6 Task- ela ed Hemodynamic Response in Visual Co ex o an Audi o y-Mo o Task ...................................................................................... 87 3.7 Discussion ........................................................................................... 94 4 Slow d i s in b ain blood olume a e associa ed wi h beha io al changes in p ima e V1....................................................................................... 99 x i 4.1 Abs ac ............................................................................................... 99 4.2 In oduc ion ....................................................................................... 100 4.3 Me hods ............................................................................................. 104 4.3.1 Summa y ................................................................................... 104 4.3.2 Su ge y, eco ding chambe s and a i icial du a .................... 105 4.3.3 Beha io ..................................................................................... 105 4.3.4 Imaging ...................................................................................... 106 4.3.5 Elec ophysiology ...................................................................... 107 4.3.6 A ousal Rela ed Index (ARI) .................................................... 108 4.3.7 Ale ness me ics: b ain mo emen and eyes open/closed ... 109 4.3.8 In e neu onal co ela ions: elec ode pai -wise co ela ions . 109 4.3.9 Co ela ions be ween pe o mance and di e en esponses 109 4.3.10 Co ela ions be ween hea a e and hemodynamics ............ 110 4.3.11 C oss- alida ion (k- old c oss- alida ion) ................................ 110 4.4 Resul s............................................................................................... 111 4.4.1 Hemodynamic mean ial esponse co ela es in e sely wi h engagemen .............................................................................................. 111 4.4.2 Complex in e play be ween hemodynamics and hea a e .. 116 4.4.3 Di e en neu al me ics and hei co ela ion o engagemen 119 4.5 Discussion ......................................................................................... 123 4.6 Acknowledgmen s ............................................................................ 127 5 Gene al Discussion .................................................................................. 129 5.1 Exogenous esponses in V1 ............................................................ 129 5.2 Endogenous esponses in V1 ......................................................... 130 5.3 Fu u e di ec ions ............................................................................... 133 5.3.1 Locus coe uleus – no ad ene gic sys em ............................... 133 5.3.2 Neu al basis o neu oimaging .................................................. 134 Re e ences ....................................................................................................... 136 1 1. In oduc ion 1.1 Hemodynamic esponses in he b ain Unde s anding he unc ioning o he b ain has cap u ed ou ascina ion o a long ime. In he ques o unde s and he inne wo kings o he b ain, se e al echniques ha e been de eloped. These ange immensely in applica ion and scope. In his hesis we used a combina ion o neu oimaging and elec ophysiology o p obe ques ions pe aining o he beha ing b ain. The space o ques ions ha can be answe ed using neu oimaging in ale subjec s is ex emely as . The ques ions add essed he e ha e as seed he esul s ound p e iously in he lab (Si o in & Das 2009), whe e a ask- ela ed hemodynamic signal was desc ibed in ale non-human p ima es, when animals we e engaged in a pe iodic ixa ion ask. This hemodynamic ask- ela ed signal was obse ed in he absence o concu en changes in local neu onal ac i i y. In his hesis we aimed a unde s anding, i s ly i he he- modynamic esponses o ex e nal s imuli ha e a comp ehensi e ela ionship o unde lying changes in spiking ac i i y. Fu he mo e, we wan ed o unde - s and he co ela ions, i any, be ween he hemodynamic esponse and he beha io o he subjec s; hus explo ing also endogenous esponses and no only esponses o ex e nal senso y s imuli. To add ess hese ques ions, we s udied ale non-human p ima es (as in he o iginal s udy, Si o in & Das 2009). In e es ingly, such signals we e also obse ed in humans (Jack e al. 2006, Syl es e e al. 2007, Donne e al. 2008). I is wo h no ing, hese eco dings in human subjec s comp ised only neu oimaging. The eco ding o elec ophysiologycal ac i i y in human subjec s ends o ely on echniques like EEG; hose we e no used in men ioned human s udies. Mo eo e , EEG eco dings migh no e lec unubiqui ous spiking ac i i y (Sal zbe g e al. 1971). Elec ophysiological 2 echniques aiming o eco d spiking ac i i y end o be s ill in usi e echniques no ou inely pe o med in human subjec s. We eco ded ac i i y om he p ima y isual co ex (V1), which is he en y poin o isual in o ma ion in o co ex. This is a s uc u e on he co ical su ace, which has been e y well cha ac e ized in e ms o i s neu al esponses. Known p ope ies o V1 da e back o he seminal wo k by Hubel and Wiesel (Hubel & Wiesel 1968). V1 has a e ino opic o ganiza ion, and no only is posi ion ela i e o he o ea encoded in i s esponses, he e a e also se e al known p ope ies o isual s imuli ha can be di e en ially encoded in V1, as con as , di ec ion, o ien a ion, spa ial o empo al equency. By combining op ical imaging o in insic signals (OIIS) wi h simul aneous mul i-uni elec ophysiological ac i i y (MUA) in V1, i is possible o add ess ques ions pe aining o he neu al unde pinnings o he neu oimaging esponse, as well as o imp o e ou unde s anding o he signals encoded in isual co ex. Finally, V1 is a co ical egion dedica ed o ision, bu ha ecei es modula- o y inpu om mul iple s uc u es in he b ain. I is known o be in luenced by endogenous a iables such as a en ion (Bashinski & Bacha ach 1980) o he iming o ewa d deli e y (Shule & Bea 2006, his is wo k wi h oden s). V1 is he e o e pa icula ly well posi ioned o s udy he con ibu- ions o endogenous and exogenous signals in o he hemodynamic espon- ses and hei neu al co ela es. 1.2 Neu oimaging: in insic signal op ical imaging In insic signal op ical imaging echniques we e de eloped in he 80's by Ami am G in ald's g oup (G in ald e al. 1986). Op ical imaging o in insic signals (OIIS) echniques measu e changes in he in ensi y o ligh e lec ed o co ical issue. I one illumina es co ex wi h isible ligh he la ges abso be o ligh in issue is hemoglobin. Thus his o m o imaging allows he moni o ing o changes in hemoglobin issue concen a ion; analogous o 3 how one can iden i y ascula u e in issue, i.e., dis inguishing a e ies and eins below he skin, by isual inspec ion. A di e en wa eleng hs, hemoglobin abso bs ligh di e en ly. Depending on hemoglobin’s edox s a e ligh is also abso bed di e en ly (e. g. Zijls a & Buu sma 1997), hence i allows o dis inguish changes in oxy-hemoglobin (HbO), deoxy-hemoglobin (HbR) and o al blood olume (HbT; he sum o oxy- and deoxy-hemoglobin). OIIS a wa eleng hs equally abso bed by oxy- and deoxy-hemoglobin (isosbes ic poin s, wa eleng hs o o e lap o he abso p ion spec a) p o ides in o ma ion abou HbT. 530 nm is an isosbes ic poin . Ano he ele an wa eleng h is a ound 605 nm; a his wa eleng h HbR abso bs ligh a ound 5 old mo e s ongly han HbO. The combined imaging a he wo wa eleng hs allows us o decompose he ne esponse in o HbO and HbR componen s (Si o in e al. 2009); o , in o blood olume and blood oxygena ion componen s. Blood oxygena ion is wha is ypically measu ed in unc ional magne ic esonance imaging ( MRI), blood-oxygen le el dependen (BOLD) (Ogawa e al. 1990). Pa en he ically, simul aneous eco ding o MRI and OIIS signals show a good ag eemen be ween he wo echniques, e.g. Jezza d e al. 1994, Fukuda e al. 2006 (in his la e s udy hey used an isosbes ic poin , wa eleng h o 570 nm). The da a p esen ed in his hesis was acqui ed a 530 nm (isosbes ic poin ; g een wa eleng h). This wa eleng h p o ides a la ge pe cen signal change han say he 605 nm (“deoxy-hemoglobin signal”), and is a signal wi h a known ela ionship be ween hemoglobin species (Si o in e al. 2009). A his wa eleng h he hemodynamic signal is a p oxy o changes in o al blood olume. When he co ical su ace is illumina ed wi h isible ligh , ligh is p e e en ial- ly abso bed by he ascula u e, as men ioned abo e. Acqui ed images a e da ke whe e ascula u e lies. Depending on image quali y and magni ica- ion one can iden i y blood essels. Capilla ies oo small o be esol ed in he images comp ise he capilla y bed and pa enchyma and hey also con ibu e o changes in he imaging signal. The e a e a numbe o o he 4 aspec s ha can con ibu e o he global OIIS, as ligh sca e ing. In he in ac b ain, hea a e o b ea hing should also be aken in o accoun , as he e is e idence o hei in luence in he BOLD signal (Chang e al. 2009, Bi n e al. 2008). By using wo di e en wa eleng hs in he same expe imen (530, and 605 nm), we can dis inguish a e ies and eins. A e ies ha e low con as when imaged a 605 nm, bu eins ha e high con as . A 530 nm, bo h a e ies and eins should ha e simila con as ; hence compa ing images acqui ed a hese wa eleng hs pe mi s dis inguishing a e ies om eins. E en hough he esul s included on his hesis e lec changes measu ed wi h he 530 nm illumina ion wa eleng h, some measu es we e made using bo h 530 and 605 nm. The in ensi y alue esul ing om he OIIS (ligh in ensi y om a CCD came a) is in a bi a y uni s; i s absolu e alue is no in o ma i e, bu di e en ial analysis can in o m abou changes in he hemodynamic species’ concen a ion. Bonhoe e and G in ald (Bonhoe e & G in ald 1991) p esen ed sub ac - ion and di ision me hods o analysis o his ype o da a; hey ound simila esul s wi h ei he app oach. In he expe imen s included in his hesis we used ac ional changes o he ligh in ensi y e lec ed o he co ical su ace, ela i e o he mean (ligh e lec ed), as a p oxy o changes in concen a ion o o al blood olume. In Chap e 2 we used ac ional chan- ges in he in ensi y o ligh e lec ed o he co ical su ace, he e inc eases in blood olume a e e lec ed as dec eases in he ac ional changes, bu Chap e s 3 and 4 use he nega i e o his change, in his case inc eases in blood olume a e e lec ed as inc eases in he signal’s ac ional change ( his di e ence e lec s publica ion his o y). The expe imen al se up was composed o a CCD came a, an ensemble o lenses ocusing he imaging plane in he co ical su ace, aking pic u es o he b ain su ace. A ached o he eco ding chambe is he illumina o (a ing like s uc u e holding op ical ibe s; each ibe is coupled o a LED) as 5 well as he elec ode holde (ano he ing like s uc u e wi h he capaci y o hold one o wo elec odes connec ed o elec ode mic o d i e s). The imaging egion is kep unde s e ile condi ions and sepa a ed om con ac wi h ou side ai (a i anium chambe encloses a clea glass, sealing he chambe ). To ensu e op ical cla i y he imaging chambe i is illed wi h a clea solu ion o aga ose. A schema ic o he se up can be ound in Figu e 1.1. In Figu e 1.1a, one can see he edge o he imaging lenses, as well as he elec ode holde (me al piece a ached o eco ding chambe ). The illumina ion sys em is composed o op ical ibe s coupled o LEDs, i p odu- ces a uni o m illumina ion p o ile as in Figu e 1.1a. In Figu e 1.1b is a close- up pic u e o he eco ding chambe , whe e can be no iced an ape u e ( h ough a silicon-coa ed glass, sealing he chambe ) o one eco ding elec ode. Figu e 1.1 – The expe imen al se up. (a) Pic u e o he se up illus a ing: imaging a angemen , wi h LED illumina ion o 530nm. (b). Pic u e o he imaging egion wi h single elec ode. Images we e acqui ed a 15 Hz (o 7.5 Hz, i wo wa eleng hs we e used). Below is a ypical image (Figu e 1.2a, b) om one o ou expe imen s. He e i can be no iced he a e y ha comes on g ay on Figu e 1.2a and is da k in Figu e 1.2b, has a di e en p o ile om eins ( isible and da k on bo h images). When looking a he ime se ies, a his imaging equency, one can clea ly no ice he e ec o he b ain’s pulsa ion (hea bea d i en a b 6 palpi a ion) (Figu e 1.2d). Pulsa ion can be emo ed by sub ac ing a high- pass il e ed e sion o he imaging signal (Figu e 1.2c). F om his pulsa ion one can es ima e hea a e. The use o neu oimaging echniques ul ima ely aims a es ima ing neu al ac i i y, bu nei he OIIS no BOLD MRI, a e di ec measu es o neu al ac i i y ( e iew on BOLD in e p e a ion, e.g. Logo he is & Wandell 2004). Figu e 1.2 In insic signal op ical imaging in V1. (a) Re lec ance o co ical su ace wi h inciden ligh o 605 nm. (b) Simila o a, bu inciden ligh o 530 nm. The le e s in ed: a – a e y, – ein. (c) A e age e lec ance o he i s 30 s o imaging (wa eleng h: 530 nm), in g ay is aw signal and in g een is he signal wi hou he high equency pulsa ion (low-pass il e ed e sion o he aw signal). (d) Pulsa ion signal obse ed a 530 nm on he i s 30 s o imaging. a b c d 010 20 30 2.9 3 3.1x104 ime (sec) e lec ance (a.u.) 010 20 30 -400 0 400 ime (sec) e lec ance (a.u.) a a 7 1.3 Neu al basis o in insic signal op ical imaging In his sec ion, I aim o b ing a en ion o he complexi y o he neu oimaging signals and hei espec i e in e p e a ions. Spiking ac i i y is an ene ge ically demanding p ocess o he b ain, wi h se e al componen s o neu al ac i i y weighing in (Laughlin e al. 1998, A well & Laughlin 2001). Me abolic demand in he b ain is mos ly epleni- shed by he blood supply, as neu ons do no ha e signi ican ene gy ese es. The e o e, i he ela ionship be ween neu onal ac i i y and me abolic demand, and be ween me abolic demand and blood supply was linea , neu oimaging would ha e a clea ela ionship wi h unde lying neu onal ac i i y. Howe e , he e is ample e idence o a mo e complex ela ionship. Thus, unde s anding he neu o- ascula coupling, he coupling o spiking ac i i y o me abolism and om me abolism o blood supply, will help in e p e neu oimaging. The main me abolic subs a es o he b ain a e glucose and oxygen; bo h each he b ain h ough blood ci cula ion. The e is conside able esea ch in o he me abolic cos o spiking ( o es ima ions o me abolic neu al cos in he oden b ain see: A well & Laughlin 2001). The de elopmen o au o- adiog aphy allowed measu ing local glucose consump ion in ale and anes he ized oden s using a adioac i e analogue o glucose, which pe mi s moni o ing o me abolic a e (Sokolo e al. 1977). The de elopmen o posi on emission omog aphy pe mi ed a be e empo al and spa ial mapping o adioac i e species (a b oad pe spec i e o he ield is p esen ed in Raichle 1979). Mo e ele an o he wo k p esen ed in his hesis: me abolic measu es ha e been used in p ima y isual co ex o ale humans unde going physiological isual s imula ion. These con i med an inc eased me abolic a e wi h isual simula ion, using nuclea magne ic esonance spec oscopy (P icha d e al. 1991). Finally when conside ing oxygen consump ion, he e ha e been epo ed inc eases in oxygen a e o 8 consump ion in esponse o elec ic pulses (in abbi s’ isola ed neu ons, Ri chie 1967). Bo h glucose and oxygen consump ion inc ease in esponse o neu onal ac i a ion, he e o e measu es o me abolism can be app op ia e p oxies o moni o ing changes in neu al ac i i y. Ul ima ely, i hemodynamic changes e lec me abolism, hen he hemodynamic esponse can be an a ac i e p oxy o changes in neu al ac i i y. Gi en he ad en o echniques, such as MRI ha allows moni o ing la ge po ions o b ain ( i ually non-in usi ely) make hose inc easingly ele an . I should be no ed ha OIIS equi es he imaged su ace o be exposed, making i an in usi e echnique. The use o neu oimaging echniques should howe e be combined wi h esea ch e o s on i s in e p e a ion. In insic signal op ical imaging also e lec s ligh sca e ing om neu opil ac i a ion. Ea ly wo k in isola ed ne e cells showed ha ac ion po en ials in unmyelina ed isola ed ibe s cause bi e ingence changes as well as ligh sca e ing, in he isible ligh spec um (Cohen e al. 1968). The e is also e idence om wo k done in hippocampal slices, whe e ansmi ed and e lec ed ligh p o iles we e obse ed o be, on a e age, symme ic and opposi e in sign. This sugges s a bigge ole o ligh sca e ing han ligh abso p ion wi h in insic signal op ical imaging (Ai ken e al. 1999). Wo k wi h in insic signal op ical imaging in he ol ac o y bulb o anes he ized oden s used mul iple imaging wa eleng hs o add ess whe he he esponse o s imula ing odo s in ol ac o y glome uli esul s mo e om ligh sca e ing in issue wi h neu onal ac i a ion o om changes in HbO/HbR concen a ion in blood. In hei case, hey concluded ha ligh sca e ing is in he o igin o he imaging esponse (Meis e & Bonhoe e 2001). Neu onal ac i i y is esponsible o o he changes ha can be accessed by looking a hemodynamic esponses, o he han di ec ligh sca e ing changes Func ional hype emia – inc eased blood low in esponse o neu onal ac i i y – also con ibu es o hemodynamic changes. As ocy es ha ha e a p i ileged ela ionships wi h blood essels a e known o play an impo an ole in media ing hype emia ( o a e iew see Iadecola & 15 In ou ea lie wo k (Si o in & Das 2009), we only compa ed b ain signals a he wo ex emes o isual d i e. To measu e s imulus-e oked signals, we used nea -maximal s imulus in ensi ies a which he isual inpu domina ed; meanwhile, we cha ac e ized he ial- ela ed signal only in essen ially comple e da kness. A ques ion no explo ed in he ea lie wo k was how hese signals would in e ac when p esen ed oge he in di e en p opo - ions in ou ine isual asks in ol ing s imuli o a ied in ensi ies, and how his admix u e o signals would a ec he in e p e a ion o b ain images. We add essed hese ques ions using ou echnique o simul aneous op ical imaging and elec ode eco ding in ale , ask-engaged macaques. He e, howe e , we p esen ed isual s imuli o e he ull con as ange (0% o 100%); o some expe imen s, we also included ials in comple e da kness. This allowed us o es whe he he ne imaging signal could be sepa a ed in o s imulus-e oked ( ha is, co ela ed wi h s imulus con as and e oked neu al spiking) and ial- ela ed componen s (dependen on ask s uc u e, bu no s imula ion o local spiking) o e a ull ange o V1 spiking and hemodynamics. Fu he mo e, as he p ima y use o neu oimaging is o es ima e local neu al ac i i y (o en done implici ly, bu also quan i a i ely by decon ol ing he imaging signal using an HRF (Glo e 1999)), we examined he accu acy o his es ima e wi h and wi hou co ec ing o he ial- ela ed signal. 2.3 Me hods 2.3.1 Summa y Simul aneous in insic-signal op ical imaging and elec ophysiology we e acqui ed om ale macaques engaged in passi e ixa ion asks (n = 34 si es, 5 hemisphe es in 3 monkeys, plus 14 addi ional expe imen s) using me hods de eloped p e iously (Bonhoe e & G in ald 1996, Si o in & Das 2009; Sh oye man e al. 2000). All expe imen al p ocedu es we e pe o med 16 in acco dance wi h he US Na ional Ins i u es o Heal h Guide o he Ca e and Use o Labo a o y Animals and we e app o ed by he Ins i u ional Animal Ca e and Use Commi ees o Columbia Uni e si y and he New Yo k S a e Psychia ic Ins i u e. 2.3.2 Beha io and s imuli Animals held ixa ion pe iodically o juice ewa d, cued by he colo o a ixa ion spo ( ixa ion window, 1.0–3.5 deg ees in diame e ; moni o dis an- ce, 133 cm; ixa ion du a ion, 3–4 s; ial du a ion, 10–20 s). Fo expe imen s wi h isual s imula ion, s imuli consis ed o sine-wa e g a ings (con as s, 0% (blank), 6.25%, 12.5%, 25%, 50% and 100%; mean luminance = back- g ound luminance = 46 cd m−2; spa ial equency, 2 cycles pe deg ee; d i speed, 4 deg ees pe s; diame e , 2–4 deg ees; o ien a ion op imized o he elec ode eco ding si e). T ials ypically comp ised single ixa ions, wi h s imulus p esen ed du ing ixa ion. Fo some expe imen s (Figu e 2.5), ials comp ised sequences o wo o h ee ixa ions wi h he s imulus p esen ed only on he i s ixa ion. S imuli we e block andomized, ha is, p esen ed in blocks each con aining a single ull se o con as s in andom o de . In he block, s imuli we e epea ed ollowing e o s (inco ec ixa ion) un il he animal had a co ec ial o each s imulus in a block ( o mul i- ixa ion ials, all ixa ions had o be co ec o a ial o be co ec ). Some expe imen s included 3.125% con as o ine esolu ion a low con as s; some o he s used a educed se o con as s o inc ease he numbe o ials pe condi- ion. Da k- oom expe imen s we e pe o med in a comple ely da k oom wi h he moni o co e ed and he ixa ion poin behind a pinhole (as desc ibed in e . Si o in & Das 2009). Eye ixa ion and pupil diame e we e eco ded using an in a ed eye acke (Ma suda e al. 2000). 17 2.3.3 Su ge y, eco ding chambe s and a i icial du a A e he monkeys we e ained on isual ixa ion asks, c anio omies we e pe o med o e he animals’ V1 and glass-windowed s ainless s eel eco - ding chambe s we e implan ed, unde su gical anes hesia, using s anda d s e ile p ocedu es (Sh oye man e al. 2000), o image a ~ 79mm2 a ea o V1 co e ing isual eccen ici ies om ~1 o 5°. The exposed du a was esec ed and eplaced wi h a so , clea silicone a i icial du a. A e he animals had eco e ed om su ge y, hei V1 was op ically imaged, ou inely, while hey engaged in he ixa ion ask. Reco ding chambe s and a i icial du a we e ab ica ed in ou labo a o y using published me hods (A ieli e al. 2002). 2.3.4 Ha dwa e Came a, Dalsa 1M30P (binned o 256 × 256 pixels, 7.5 o 15 ames pe s); ame g abbe , Op ical PCI Bus Digi al (Co eco Imaging). So wa e was de eloped in ou labo a o y based on a p e iously desc ibed sys em (Kala sky & S yke 2003). Illumina ion, high-in ensi y LEDs (Agilen Techno- logies, Pu dy Technologies) wi h emission wa eleng h cen e ed a 530 nm (g een, equally abso bed in oxy- and deoxyhemoglobin). Lens, mac oscope o back- o-back came a lenses ocused on he co ical su ace. Imaging, ial da a ( ial onse , s imulus onse , iden i y and du a ion, e c.) and beha io al da a (eye posi ion, pupil size, iming o ixa ion b eaks, ixa ion acquisi ions, ial ou come) we e acqui ed con inuously. Da a analyses we e pe o med o line using cus om so wa e in MATLAB (Ma hWo ks). 2.3.5 Image p e-p ocessing P io o analysis, acqui ed images we e (i necessa y) mo ion co ec ed by aligning each ame o he i s ame by shi ing and o a ing he images using he blood essels as a e e ence (Lucas & Kanade 1981). Slow empo al d i s (>30 s) we e emo ed wi h high-pass il e ing, and co ical pulsa ions wi h low-pass il e ing using he Ch onux MATLAB Toolbox 18 unc ion unline.m ( ypical hea a es we e ~2–3 Hz, much as e han he ypical hemodynamic esponse equencies o ~<0.5 Hz). 2.3.6 Elec ophysiology Elec ode eco dings we e made simul aneously wi h op ical imaging. Reco ding elec odes (FHC, AlphaOmega; ypical impedances we e ~600– 1,000 kΩ) we e ad anced in o he eco ding chambe h ough a silicone- co e ed hole in he ex e nal glass window, using a cus om-made low-p o ile mic od i e. Reco ding si es we e mos ly, bu no exclusi ely, con ined o uppe laye s. Signals we e eco ded and ampli ied using a Plexon eco ding sys em. The elec ode signal was spli in o spiking (100 Hz o 8 kHz band- pass) and LFP (0.7–170 Hz); LFP da a no shown. No a emp was made a isola ing single uni s and all measu ed spiking was MUA (de ined as each nega i e-going c ossing o a h eshold = ~4× he .m.s. o he baseline ob ained while he animal looked a a g ey sc een (Si o in & Das 2009)). The MUA signals we e hen high-pass il e ed o emo e slow d i s (>30 s), down sampled o he imaging ame a e (7.5 o 15 samples pe s) and aligned o line wi h he images. 2.3.7 HRF ke nel i ing Each HRF was modeled as a gamma- a ia e unc ion ke nel o he o m whe e  = (T/W)2 *8.0*log(2.0),  =W2/(T*8.0*log(2.0)), A is he ampli ude, T is he ime o peak and W is he ull wid h a hal maximum (Cohen 1997; Si o in & Das 2009; Madsen 1992). This unc ional o m allows o pa ame ically a ying ke nel ampli ude, la ency and wid h. Fo i ing, we used a downhill simplex algo i hm ( minsea ch, MATLAB) minimizing he sum squa e di e ence be ween measu ed and p edic ed hemodynamics. All i s used pe iodic unc ions (30 epe i ions) cons uc ed om means o he 19 ele an signals ac oss con as s, aligned o ial onse s, o co ec ials alone. Thus, HRFSTIM was ob ained by i ing a pe iodic pa e n o he mean SSTIM o he mean HSTIM, he HRFNULL by i ing he mean S o he mean H (o e co ec ials alone), he HRFBLANK by i ing he mean SBLANK o he mean HBLANK, and he HRFDARK by i ing he mean SDARK o he mean HDARK. 2.3.8 Goodness o i o p edic ed hemodynamics Fi was quan i ied as R2 = 1 − ( a iance o he esidual e o )/( a iance o measu ed hemodynamics) (Supplemen a y No e, equa ion (10)), exp essed ei he sepa a ely o each con as o as mean R2, ha is, calcula ed o he mean signals a e aged ac oss all con as s. Fo all i s o he han o he blank signal, p edic ions (and esidual e o s) we e calcula ed by con ol ing he ull aw measu ed spike ace wi h he ele an HRF and hen sepa a ing la e in o co ec ials by con as , o a e aging ac oss con as s. This is mo e eliable han con ol ing syn he ic pe iodic unc ions cons uc ed om mean signals because wi h pe iodic unc ions he e is a isk o ge ing a ma ch, no wi h he ue signal, bu wi h a signal phase-shi ed by a ac ion o a ial pe iod (Das & Si o in 2011). Such misma ches a e highly unlikely in he measu ed signal wi h i s andom sequence o s imulus in ensi ies and co esponding e oked hemodynamics (Das & Si o in 2011). The blank signal i using HRFBLANK was es ed using pe iodic unc ions, as in his case we we e es ing he i using a ke nel ha speci ically did no i he ull s imula ed spike sequence. 2.3.9 Boo s apping o ge con idence limi s on R2 Fo each expe imen , 200 boo s ap da a se s we e cons uc ed, each wi h he same numbe o ials as he o iginal, using andom esampling wi h eplacemen (Supplemen a y Figu e 2.10). The esampled hemodynamic and spike ials we e hen i ed agains each o he sepa a ely o bo h models (MLM and null) and R2 alues we e calcula ed as be o e (Supple- 20 men a y No e, equa ion (10)). The 95% con idence limi s we e ob ained by aking he 2.5 h o he 97.5 h pe cen iles; simila ly, 80% con idence limi s by aking he 10 h o he 90 h pe cen ile. Random eselec ion was done sepa a- ely by con as o ha e he same numbe o ials pe con as . Howe e , each con as used he same andom numbe se o main ain s imulus blocks and educe a iabili y esul ing om long- e m d i s in physiology o eco ding s abili y. This was pa icula ly necessa y o he MLM, which in ol es sub ac ing he mean blank signals HBLANK and SBLANK om all o he con as s; i blocks a e no main ained, his sub ac ion leads o a numbe o noisy ou lie s in he boo s ap es ima e when a se o blanks ials domina ed by one epoch o a session ( o example, high signal) is sub ac ed om nonblank ials domina ed by a di e en epoch ( o example, low signal). 2.3.10 C oss- alida ion o HRFSTIM ke nels ac oss sessions Fo each session, we c ea ed a lea e-one-ou mean HRFSTIM ke nel by a e aging he wo iming pa ame e s (peak la ency and wid h) ac oss all ke nels excluding he gi en one. Ke nel ampli ude was ob ained by i ing, using his mean ke nel o i he gi en session’s da a (HRFSTIM ampli ude depends on an a bi a y scale ac o in elec ode eco ding; Supplemen a y Figu e 2.11). This lea e-one-ou mean ke nel wi h he bes i ed ampli ude was hen used o ob ain he c oss- alida ion p edic ion and co esponding R2. C oss- alida ion was pe o med ei he ac oss all animals o es ic ing he lea e-one-ou a e aging o o he ke nels o he gi en animal. 2.3.11 Decon olu ion The spike ace es ima ed by decon olu ion was de ined as whe e H is he ele an hemodynamic signal, HRF is he co esponding op imal HRF ke nel and F and F−1 indica e o wa d and in e se ( as ) Fou ie 21 ans o ms, espec i ely. Gi en ha F(HRF) has low powe a high equencies, e lec ing he slow hemodynamic esponse, we il e ed using a Hamming window wi h a 0.5-Hz cu o in equency space. This a oided high- equency noise in he hemodynamic signal om being ampli ied du ing decon olu ion. The same il e was used o discoun high equencies in he measu ed spike a e be o e co ela ing wi h he decon ol ed es ima e. 2.3.12 Checking he s abili y o ou p ima y indings agains a iabili y in elec ode eco dings I measu ed spiking S is a e idical scaled sample o he ue spiking s o ou models despi e measu emen a iabili y ac oss expe imen s (di e en elec odes, di e en h esholds o spike de ec ion o MUA), hen he ampli ude o he i ed HRF should simply scale in e sely wi h measu ed spiking o a gi en expe imen (Supplemen a y Figu e 2.11 and Supplemen a y No e, equa ion (7)) The scale ac o (be ween he measu ed S and he ue s) will cancel ou in all equa ions o a gi en expe imen , lea ing model ea u es unchanged ( ha is, ke nel shape, ial- ela ed signal T and R2). We es ed o his in wo ways. Fi s , we es ed he e ec o a ying spike de ec ion h esholds. In i e expe imen s, we eco ded he elec ode signal a a low h eshold and hen e h esholded o -line o gene a e mul iple se s o spiking da a S o he same imaging da a ( o example, peak spike a es om abou 300 s−1 o abou 10 s−1 o p og essi ely highe h esholds; Supplemen a y Figu e 2.11a,b). These e h esholded spike da a we e hen i ed sepa a ely agains he common imaging signal (Supplemen a y Figu e 2.11c–g). In a second es , we checked he linea i y o he ela ion linking HRFSTIM ampli ude agains he in e se o he SSTIM ampli ude o e ou ull da a se (in eg a ion window o mean SSTIM coex ensi e wi h s imulus du a ion as in Figu e 2.2a; Supplemen a y Figu e 2.11h). 22 2.4 Resul s Fo hese expe imen s, we used h ee hesus macaques (monkeys Y, T and S; n = 34 eco ding si es ac oss i e hemisphe es; monkey S was also used p e iously (Si o in & Das 2009)). The animals’ ask, which was cued by he colo o a ixa ion spo , in ol ed ixa ing and elaxing ( ha is, ee iewing) pe iodically o a juice ewa d. This ask is known o e oke obus ial- ela ed signals in V1 ( e . Si o in & Das 2009). A ial ypically comp ised a single ixa ion, wi h ixed ial pe iodici y o 10–20 s. Fo one se o expe imen s, ials consis ed o sequences o wo o h ee ixa ions, each ewa ded o co ec ixa ion. Visual s imuli comp ised d i ing sine-wa e g a ings ha we e p esen ed passi ely while he animal ixa ed. The g a ing con as was ypically a ied in i e log2 s eps plus a blank, p esen ed in andomized o de ; he con as s a ied in some expe imen s and g a ing o ien a ion was op imized o each elec ode eco ding si e. In addi ion, o compa e wi h ou ea lie esul s (Si o in & Das 2009), we pe o med a se o expe imen s in da kness (see Online Me hods). We eco ded concu en MUA and hemodynamics om V1. Fo hemodyna- mics, we used in insic-signal op ical imaging, a high- esolu ion op ical analog o MRI ha deduces co ical hemodynamics by measu ing ac ional changes in he in ensi y o ligh e lec ed o he co ical su ace a wa elen- g hs abso bed by hemoglobin (Bonhoe e & G in ald 1996 Si o in & Das 2009; Si o in e al. 2009). We speci ically used he blood olume signal imaged a 530 nm (g een), as i di ec ly measu es changes in o al local issue hemoglobin concen a ion, and hus in local blood olume (Si o in & Das 2009; De o e al. 2003; She h e al. 2004; Nemo o e al. 2004). Fu - he mo e, i ma ches co esponding MRI signals (Fukuda e al. 2006). A pa icula ad an age o his imaging signal is ha i s impulse esponse o a b ie senso y s imulus is monophasic, wi h an inc ease in abso p ion ol- lowed by a mono onic e u n o baseline, p esumably e lec ing he s imulus- igge ed inc ease and subsequen decline in local blood olume (Si o in e 23 al. 2009; De o e al. 2003; She h e al. 2004; Nemo o e al. 2004). The monophasic s imulus- igge ed esponse makes he imaging signal easy o in e p e and o model ma hema ically (see Supplemen a y No e). 2.4.1 Spikes poo ly p edic hemodynamics in pe iodic ask Ou eco dings showed, as expec ed, s imulus-d i en spiking and hemody- namic esponses wi h ampli udes mono onically e lec ing s imulus con as ial by ial (Figu e 2.1). In addi ion, many eco dings e ealed a obus spiking signal locked o ial onse ha was common o all o he spike aces and was mos e iden o blank ials (SBLANK; Figu e 2.1b). Addi ional e i- dence sugges s, howe e , ha his blank- ial spiking, main ained low du ing ixa ion and high in be ween ixa ions, is also isual (Supplemen a y Figu e 2.8). The ime cou se o his signal ma ched ha o he animal’s eye aces (Supplemen a y Figu e 2.8a), and i was ex inguished in he da k, e en when he animal’s eye ace pa e ns emained unchanged (Supplemen a y Figu e 2.8b). This signal was he e o e likely a esul o he animal looking a ound he dimly li oom and hen a he g ay moni o , pe iodically, in each ial. 24 Figu e 2.1 – The ull hemodynamic signal is poo ly p edic ed by local mul i- uni spiking. (a) Top, a sec ion o he ull eco ded spiking signal (S) o a ep esen a i e expe imen al session (black ace). Bo om, co esponding measu ed hemodynamic signal (H, black) and he bes p edic ion ob ained om spiking (o ange, S, indica es con olu ion; Supplemen a y No e, equa ion (3)). Inse , bes i ing ke nel HRFNULL (ampli- ude no malized) ob ained by i ing H o S. Mean R2 = 0.49, as calcula ed using mean signals a e aged ac oss con as s (n = 261 ials o al, oughly 43 pe con as ). Red line segmen s indica e s imulus applica ion and e ical do ed lines indica e ixa ion ial onse . Red and black a ows below aces indica e ypical esponses o high-con as (100% con as ) and blank (0% con as ) s imuli, espec i ely; o hemodynamics, inc easing nega i e ampli udes, ha is, inc easing abso p ion o ligh by co ex, equals inc easing blood olume. No e he poo ma ch be ween he obse ed and p edic ed aces leading o a la ge esidual and, consequen ly, low mean R2. (b) T ial-aligned a e ages o spiking (S) o each con as . The ial s uc u e is indica ed by he colo ba s (g ay, ixa e; ed, s imulus; no ba , elax). No e he p ominen blank- ial spiking signal SBLANK. (c) Da a p esen ed as in b o hemodynamics (H). (d) Da a a e p esen ed as in b o co esponding p edic ed hemodynamics (solid lines, op) and esiduals (H- , do ed lines, bo om; sepa a ed e ically o isibili y). Indi idual R2, calcula ed sepa a ely pe con as , a e shown alongside each p edic ion. Da a we e ob ained om monkey S. E o ba s ep esen s.e.m. b. Time (sec) 0 2 4 6 8 10 c. dR/R 0.02 -0.04 0 -0.02 Time (sec) 0 2 4 6 8 10 Blank (0.0) 6.25 12.50 25.00 50.00 100.00 Spike a e (Hz) 0 100 200 300 Con as (%) Spikes , all co ec ials SHemo , all co ec ials H ial sequence: g ey: ' ixa e', ed: ‘s im’; es o ial: ‘ elax; d. 0 Time (sec) 0 2 4 6 8 10 dR/R 0 -0.02 a. 800 850 900 950 1000 -0.04 0 0.04 0 200 400 Spike a e (Hz) dR/R Time (sec) 0 -1 0 15 R2 = 0.49 HRFN ULL P edic ed Residual -4.21 -0.12 0.26 0.46 0.59 0.58 SBLANK HBLANK 31 esponses, a e aged o e he window as in Figu e 2.2b. (c) The spiking and hemodynamic esponses shown in a and b, plo ed agains each o he . The g ay lines a e eg ession lines o each session and he ed line is he a e age o he eg ession lines. Exp essions show eg ession and R2, bo h o he expe imen s in Figu e 2.1 and Figu e 2.2a–c (exp) and he popula ion (a g). Popula ion a e ages we e calcula ed om session alues, weigh ed by numbe o ials wi hin a session (n = 34 sessions, 3 monkeys). 2.4.3 T ial- ela ed signal consis en in s imulus and da k oom Wi h he s imulus-e oked po ion o he signal well cha ac e ized by ou MLM model, we nex es ima ed he posi ed spike- and s imulus-independen ial- ela ed signal T (Figu e 2.4and Supplemen a y No e, equa ion (2,9)). Acco ding o he MLM, his is he signal ha emains a e sub ac ing away, om he ull measu ed hemodynamics H, all componen s ha can be p e- dic ed om spikes. To es ima e spike-p edic ed componen s, we used ou simpli ying assump ion ha he HRFSTIM ke nel can be applied uni o mly o all spiking, whe he s imulus e oked o uncon olled (blank ial). The HRFSTIM was ob ained, as be o e (Supplemen a y No e, equa ion (7)), by i ing s imulus-e oked spiking (SSTIM; Figu e 2.4a) o hemodynamics (HSTIM; Figu e 2.4b). The hemodynamics p edic ed om ull spiking S using his ke nel (Figu e 2.4c) we e clea ly di e en om he ull measu ed hemody- namics (Figu e 2.4b). Quali a i ely, howe e , he la e appea o be a sum o he p edic ion iding on op o a la ge con as -independen esponse. Indeed, sub ac ing he p edic ed om he measu ed hemodynamics le la ge emaining signals T ha ma ched each o he closely ac oss con as s. No e, mo eo e , hei subs an ial s eng h, which was 1.5- old g ea e han ha o he maximal HSTIM (compa e Figu e 2.4b wi h Figu e 2.4d). I is impo an o emphasize ha , in ou amewo k, hese unp edic ed hemody- namic signals a e no he esul s o nonspeci ic spiking ( o example, SBLANK); hey comp ise he componen s ha , acco ding o he MLM, emain a e using HRFSTIM o accoun o he en i e y o he spiking- ela ed 32 hemodynamics, bo h s imulus e oked and nonspeci ic (Supplemen a y No e, equa ion (9)). Figu e 2.4 – Es ima ed ial- ela ed signal T is consis en ac oss con as s, ac oss expe imen s, and be ween s imula ed and da k- oom ials. (a,b) Spiking (a) and hemodynamics (b) om ano he ep esen a i e session. Inse s, co esponding SSTIM and HSTIM. The ial s uc u e is indica ed by he a. b. Spike Ra e (Hz) 0 200 400 600 0.02 -0.03 -0.02 -0.01 0.01 dR/R 0 0.02 -0.03 -0.02 -0.01 0.01 dR/R 0 0.02 -0.02 -0.01 0.01 dR/R 0 0.02 -0.03 -0.02 -0.01 0.01 dR/R 0 Time (sec) 0 5 10 15 Blank 3.125 6.25 12.5 25 50 100 -10 0 15 R2 = HRFST IM . c. d. e. Spike 0 200 T: S imulus T: Da k 0.02 -0.03 -0.02 -0.01 0.01 dR/R 0 Time (sec) 0 5 10 15 g. -1 -0.5 0 0.5 1 3 6 0 T: S imulus T: Da k h. 600 0 015 Co . 0 -0.02 015 Popula ion: Median Co . 0 0.25 0.5 0.75 1 0 15 30 0.02 -0.03 -0.02 -0.01 0.01 dR/R 0 0.90 Measu ed: mean P ed om spikes Spikes SDARK T ial- ela ed signals : T Hemo: H Spikes: SSTIM ial sequence: g ey: ' ixa e' S imula ed ials Da k- oom and compa ison wi h s imula ed ials Spikes: S Hemo: HSTIM Hemo HDARK Measu ed: indi idual P eds X HRFSTIM S Pai wise co ela ion T: Da k s. S imulus Popula ion T: Da k s. S imulus: T: Da k s. S imulus Co . = 0.94 # sessions # sessions 33 colo ba s (g ay, ixa e; ed, s imulus; no ba , elax). (c) P edic ion con ol ing op imal ke nel HRFSTIM (inse ) wi h ull spiking S ( ha is, HRFSTIM ⊗S). (d) Es ima ed ial- ela ed signals ( ; Supplemen- a y No e, equa ion (9)) shown indi idually by con as ; ed indica es mean ac oss con as s. No e he high ampli ude o mean T (s.d. = 0.0088, compa ed wi h 0.0058 o HSTIM using 100% con as ; a 1.5- old di e ence). No e he ma ked simila i y o signals T ac oss con as s (co ela ion (Pea son’s ) wi h lea e-one-ou means: 0.99, 0.99, 0.99, 0.99, 0.99, 0.99 and 0.96 o con as s 0–100% in sequence; n = 175 ials, 25 pe con as , 7 con as s, median = 0.99). Inse , popula ion his og am o median (mean (s.e.m.) = 0.94 (0.01), n = 34). (e) Da k- oom ials o he sessions shown in a–d. Top, hemodynamic aces, HDARK. G ay lines a e indi idual aces, all co ec ials (n = 45), he g een line is he mean o co ec ials, and he ed line is he p edic ion, con ol ing HRFSTIM wi h da k- oom spiking SDARK (bo om ace, black). T ial s uc u e indica ed on ime axis (pe iodic ixa ions in da kness). ( ) Mean ial- ela ed signals T om s imulus-d i en (black) and da k- oom ials (g een), same session (Pea son’s = 0.94). (g) Signals T as in o ull popula ion (n = 19 pai s, monkey T). (h) Pai wise co ela ions be ween s imula ed and co esponding da k- oom T o e popula ion (mean (s.e.m.), pai wise Pea son’s = 0.61 (0.08), n = 19 pai s). We quan i ied he simila i y o he ial- ela ed signals T o each o he , a di e en con as s in an expe imen , by co ela ing he signal T a each con as (including con as = 0, blank) wi h he lea e-one-ou mean o he signals T calcula ed a all he o he con as s. All o he esul an co ela ion (Pea son’s ) alues we e e y close o 1.0 (Figu e 2.4d). This pa e n was epea ed o e ou popula ion o 34 expe imen s gi ing, in each case, a median close o 1.0 (Figu e 2.4d). We wan ed o es how well he ial- ela ed signals hus calcula ed ma ched each o he ac oss expe imen s and how simila hey we e o he ial- ela ed signals obse ed in da k- oom ixa ion asks (Si o in & Das 2009). Fo monkey T, we we e success ul in ge ing se s o bo h da k oom and isually s imula ed ials in 19 expe imen s (5 o he cu en 34, and an addi ional 14 om a sepa a e p ojec using he same ixa ion ask). O e his popula ion, we ound a close ma ch o each esidual wi h he da k- oom signal a he 34 same eco ding si e, as well as a ma ked simila i y o hese signals ac oss expe imen s (Figu e 2.4e–h). As in ou ea lie published da a (Si o in & Das 2009), he da k- oom ials e oked high-ampli ude s e eo yped signals o clockwo k-like pe iodici y despi e weak spiking (Figu e 2.5e). Howe e , he da k- oom ial- ela ed signal T ( ha is, a e sub ac ing he, albei e y small, spike- ela ed p edic ion ob ained by con ol ing wi h HRFSTIM; Figu e 2.4e) closely ma ched he mean ial- ela ed signal T om he isually e oked ials (Figu e 2.4 ). A simila pa e n was seen o each expe imen . The se s o all da k- oom and isually s imula ed ial- ela ed signals we e ma kedly simila (Figu e 2.4g) and ma ched each o he well when co ela ed pai wise o each eco ding si e (Figu e 2.4h). This p o ides compelling e idence o ou MLM, ha is, ha he ull hemodynamic signal e oked in an ale ask-engaged subjec is he linea sum o a spike- ela ed componen and a dis inc ial- ela ed componen ha is independen o local spiking o isual s imula ion (Supplemen a y No e, equa ion (2)). As an addi ional es o ou p emise ha he ial- ela ed signal is de e mined by ial s uc u e independen o s imulus o e oked spiking, we designed a se o expe imen s in which we a ied ial s uc u e while keeping he s imu- la ion pa ame e s unchanged (n = 10 expe imen s in 2 animals, monkeys S and T; Figu e 2.5). Bo h se ies consis ed o 30-s ials wi h iden ical s imula- ion (g a ings, wi h con as s: 0% (blank), 12.5% o 100% in block- andomi- zed o de , shown once pe ial). The ials had di e en ine s uc u e, howe e . Fo one se , he monkey made wo ixa ions a 15-s in e als in each 30-s ial (Figu e 2.5a–c), whe eas in he o he se , he monkey made h ee ixa ions a 10-s in e als pe 30-s ial (Figu e 2.5d– ). The s imulus was p esen ed only du ing he i s ixa ion, whe eas subsequen ixa ions we e on o he blank moni o . No ably, we only conside ed hose ials in which he animal pe o med sequences o co ec ixa ions ex ending o e he ull 30-s ial o be co ec ials. Blank ials, co espondingly, consis ed o wo (Figu e 2.5a–c) o h ee (Figu e 2.5d– ) successi e co ec blank- moni o ixa ions s a ing wi h he blank (0% con as ) s imulus. 35 Figu e 2.5 – Es ima ed ial- ela ed signal T e lec s ial iming independen o s imulus iming o con as . (a,b) Spiking (a) and hemodynamics (b) o ials consis ing o ixa ion sequences in which he animal ixa ed wi h 15-s pe iodici y, bu he s imulus ( h ee con as s, including 0%, blank) was shown a 30-s in e als, ha is, only a he i s ixa ion o each pai . Inse s, co esponding SSTIM and HSTIM, calcula ed by sub ac ing away blank- ial signals consis ing o esponses o he pai o ixa ions on o he blank moni o s a ing wi h he blank s imulus (indica ed by blue cu es). No e monophasic s imulus-e oked HSTIM wi h no e idence o oscilla o y ebounds du ing he blank epoch. The ial s uc u e is indica ed by he colo ba s (g ay, ixa e; ed, s imulus; no ba , elax). (c) T pe con as . Inse , op imal HRFSTIM calcula ed o e 30-s ials. No e ha he signal T is close o exac ly pe iodic a he 15-s ixa ion pe iodici y, wi h iden ical ampli udes o he i s and second ixa ion pe iods independen o s imulus s eng h o e oked spikes in he i s ixa ion (co ela ion o calcula ed T ac oss s imulus con as s: 0.98 (median o pai wise co ela ions be ween each T and he lea e-one-ou mean o he o he wo), n = 74 ials o al, oughly 25 pe con as ). (d– ) Da a a e p esen ed as in a–c, wi h he same s imuli, p esen ed a he same 30-s in e als, bu wi h he monkey ixa ing e e y 10 d. Spike a e (Hz) Time (sec) 0 100 200 300 400 a. Blank 12.50% 100.00% Spike a e (Hz) Time (sec) 0 110 220 330 440 b. dR/R Time (sec) -0.04 -0.03 0 0.02 -0.02 -0.01 0.01 e. Time (sec) dR/R -0.02 -0.15 0 0.01 -0.01 -0.005 0.005 c. Time (sec) . Time (sec) -1010 0 20 R2 = 0.96 HRFST IM 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 0 5 10 15 20 25 30 -1010 0 20 R2 = 0.97 HRFST IM T ial- ela ed signals : T Two 15-sec ixa ions in one 30-sec ial Spikes: S Th ee 10-sec ixa ions in one 30-sec ial T ial- ela ed signals : T Co ela ion = 0.98 (median) Co ela ion = 0.97 (median) Spikes: SHemo: H 010 20 30 0 -200 200 400 Spikes: SSTIM 3020100 0 -0.03 -0.015 0.015 Hemo: HSTIM 010 20 30 -100 100 0 200 300 Spikes: SSTIM 010 20 30 -0.03 -0.02 -0.01 0 0.01 Hemo: HS TIM Hemo: H 36 s. The s imulus was shown only on he i s ixa ion o each iple (n = 83 ials o al, oughly 28 pe con as ). All e o ba s indica e s.e.m. E en hough he eco ded spiking S, o bo h ial s uc u es, was domina ed by s imula ion a he 30-s ial pe iodici y (Figu e 2.5a,d), he eco ded hemodynamics H showed addi ional powe ul modula ions, e en o he blank ials (Figu e 2.5b,e). No ably, his modula ion ma ched he ixa ion schedule and was hus dis inc o he wo- ixa ion e sus he h ee- ixa ion ials. The signal ampli udes in he second and hi d in e als we e no p opo ional o he ampli ude o he signal in he i s ixa ion in e al, as hey would ha e been i hey we e a esul o inging ollowing he ini ial s imula- ion. On sub ac ing away he ele an blank- ial signals, he s imulus- e oked hemodynamics HSTIM in each case showed a monophasic decline o baseline, wi h compa able ime cou ses (Figu e 2.5b,e), as would be expec- ed o a monophasic blood- olume esponse o he s imulus Si o in e al. 2009. Finally, he ial- ela ed signals T we e, in each case, pe iodic a he ial ine s uc u e wi h no appa en modula ion by he s imulus (T ob ained as abo e by sub ac ing away om each measu ed signal H componen s p edic ed om ull spiking using he ele an HRFSTIM; Figu e 2.5c, ). 2.4.4 Spikes poo ly p edic blank- ial and da k- oom signals As a coun e p oposal o ou MLM, i could be a gued ha he e is no need o in oke any special spike-independen ial- ela ed signal T. Al hough we ha e p o ided e idence ha blank sub ac ion leads o a ma kedly imp o ed linea i be ween he s imulus-e oked po ions o he signal, i does no ollow ha he blank- ial hemodynamics necessa ily con ain signal compo- nen s independen o spiking as p oposed in he MLM. Ins ead, i could be ha blank- ial hemodynamics a e ela ed linea ly o he some imes subs an- ial blank- ial spiking (Figu e 2.1b) h ough a dis inc HRF ke nel app op ia- e o low spiking le els ha is e y di e en om he ke nel linking he s i- mulus-e oked signals. This, one could a gue, is he eason o he misma ch 37 when ying o p edic he ull hemodynamics om he ull spiking (Figu e 2.1). As we show below, howe e , any such dis inc HRF ke nels appea a bi a ily a iable and un eliable, making his coun e p oposal highly nonpa simonious and hus implausible. We i s es ed whe he he blank- ial hemodynamics could be p edic ed linea ly om spiking alone (Figu e 2.6a–d and Supplemen a y No e, equa- ion (11)), as opposed o being modeled by a sum o spike-p edic ed and ial- ela ed componen s ( ha is, MLM; Supplemen a y No e, equa ion (4)). Indeed, we could make a easonable p edic ion (R2 = 0.46; Figu e 2.6a) by using he op imal blank- i ed ke nel HRFBLANK ob ained by i ing he blank- ial hemodynamics o blank- ial spiking. This was be e han he alue o R2 = −0.19 o he p edic ion using he same session’s HRFSTIM (Figu e 2.6a). This, howe e , is no su p ising. By de ini ion, he i ing p ocess disco e s a ke nel ha maximally accoun s o he a iance in he i ed signal. Howe e , he i he e was likely o ui ous. The blank- i ed ke nel was i e old la ge in ampli ude and opposi e in sign o he HRFSTIM, gi ing absu d p edic ions o he s imulus-e oked signal when con ol ed wi h he same session’s s imulus-e oked spiking SSTIM (Figu e 2.6b). O e he popu- la ion, hese blank- i ed ke nels we e highly a iable in ampli ude ela i e o he co esponding HRFSTIM (Figu e 2.6c) and, mo eo e , showed a wide sca e in peak la ency and wid h (Figu e 2.6c,d). No ably, he p esence o bo h posi i e and nega i e ampli udes made i meaningless o e en pe o m a c oss- alida ion es o see how well he ke nel om one day can be used o p edic he blank signals om o he days, in sha p con as o he eliable p edic ions ob ained h ough c oss- alida ion o he HRFSTIM (Figu e 2.2 ). 38 Figu e 2.6 – Blank- ial and da k- oom hemodynamic esponses a e poo ly i ed o spiking. (a) Blank- ial esponses (same expe imen as shown in dR/R HRFDAR K 010 -45 0 515 R2 = 0 200 0 0.02 -0.03 -0.02 -0.01 0.01 Hz e. g. HRFBLANK HRFDARK HRFDARK Da k- oom- i ed s. blank- i ed ke nels Time (sec) 0 5 10 15 0 15 Time (sec) 0 15 Time (sec) 015 Peak la ency Peak wid h Blank- i ed s. s imulus- i ed ke nels Time (sec) 0.58 Da k- oom: Spikes: SDARK Da k- oom: Hemo Measu ed: HDARK P ed: using HRFDARK Time (sec) a. Hz 0 -0.02 -0.01 0.01 dR/R 0 5 10 15 0 200 HRFB LA NK 0010 5 515 R2 = 0.46 Measu ed: HBLANK P ed: using HRFBLANK P ed: using HRFSTIM Blank: Spikes: SBLANK Blank: Hemo Peak la ency Peak wid h HRFBLANK HRFSTIM HRFSTIM 015 0 15 Time (sec) 015 d. Time (sec) Time (sec) b. Time (sec) 0 5 10 15 0.04 -0.01 0 0.01 0.03 dR/R 0.02 0.05 0.06 0 -0.02 015 S im: Hemo p ed using HRFBL ANK HSTIM -20 -10 0 10 HRFBLANK HRFDAR K no malized by HRFBLANK 200 5 10 15 -1 0 1 . Time (sec) 0 3 -20 -10 0 10 Amp. Blank/S im # sessions 20 -10 0 10 200 5 10 15 HRF HRF B LA NK STIM no malized by c. 0-10-20 10 Amp. Blank/S im 0 2 6 4 # sessions Time (sec) HRFBLANK 015 Time (sec) HRFBLANK Time (sec) HRFBLANK 015 Time (sec) HRFBLANK 015 Time (sec) HRFBLANK Time (sec) 39 Figu e 2.4). Top, SBLANK. Bo om, co esponding HBLANK compa ing he measu ed alue (blue) wi h wo al e na i e p edic ions: one om SBLANK using session’s s imulus- i ed ke nel ( ed, HRFSTIM SBLANK, R2 = −0.19, n = 25 ials) and he o he using he op imal blank- i ed ke nel (b own, HRFBLANK SBLANK, R2 = 0.46). Inse , he blank- i ed ke nel HRFBLANK (ampli ude no malized o HRFSTIM). G ay ba indica es ixa ion. (b) P edic ions (HRFBLANK SSTIM) o s imulus-e oked hemodynamics using blank- i ed ke nel. Compa e wi h measu ed HSTIM (inse ) (mean R2 = −11.9). (c) Popula ion o HRFBLANK ke nels, each no malized by ampli ude o co esponding HRFSTIM ke nel. Inse , his og am o HRFBLANK ampli udes no malized by co esponding HRFSTIM (Amp. Blank/S im., n = 34). (d) Peak la encies (le ) and wid hs ( igh ) o HRFBLANK e sus HRFSTIM. No e he high a iabili y (s.e.m.) in bo h pa ame e s o he HRFBLANK. Popula ion a e age la ency (s.e.m): blank, 4.6 (0.6) s; s imulus, 3.1 (0.2) s; popula ion a e age wid h (s.e.m.): blank, 5.8 (1.0) s; s imulus, 3.3 (0.2) s; n = 33; igno ing one ou lie wi h wid h = 1.25 × 108 s o he blank). (e) Da a a e p esen ed as in a o he da k- oom ask (da a om Figu e 2.4e). Top, SDARK. Bo om, co esponding HDARK compa ing measu ed ace (g een) wi h p edic ion using he op imal da k- i ed ke nel (magen a, HRFDARK SDARK, R2 = 0.58). Inse , da k- i ed ke nel HRFDARK (ampli ude no malized o HRFSTIM). ( ) Top, HRFBLANK ke nels no malized by absolu e alues o hei own ampli udes (no e a iabili y in ime cou ses). Bo om, HRFDARK ke nels no malized by co esponding HRFBLANK ampli udes. Inse , his og am o HRFDARK ampli u- des no malized by HRFBLANK (Amp. Da k/S im., n = 19 sessions). (g) Sca e plo s o peak la encies (le , Pea son’s = 0.04) and wid hs ( igh , =0.07) compa ing HRFDARK and co esponding HRFBLANK (n = 18; ou lie igno ed as in d). The p esence o bo h da k- oom and isually s imula ed ials o 19 eco - ding si es allowed o addi ional es s o he coun e p oposal o he MLM. I he e exis alid low-spiking-le el ke nels linea ly linking blank- ial spiking o hemodynamics, hen such ke nels should also be easonable o linking da k- oom spiking o hemodynamics. We es ed his possibili y by calcula ing he op imal da k- i ed ke nels o each o hese sessions (Supplemen a y No e, equa ion (12)). These ke nels, again p o ided, by de ini ion, good i s o he gi en da k- oom signals; bu hey we e a bi a ily di e en om he co esponding blank- i ed ke nels. Thus, o he pa icula 40 example session, he da k- i ed ke nel was opposi e in sign and much la ge in ampli ude (40- old e sus i e old la ge han he ampli ude o he session’s HRFSTIM; Figu e 2.6e), e lec ing he smalle da k- oom spiking ampli ude compa ed wi h blank- ial spiking. O e he popula ion, we ound simila ly poo co espondence in ampli ude, la ency and wid h be ween da k- and blank- i ed ke nels (Figu e 2.6 ,g). The appa en ly a bi a y shapes and sizes o hese ke nels, when combined wi h ou ea lie e idence o law ul and s e eo yped ial- ela ed signals when i ing he MLM (Figu e 2.4 and Figu e 2.5), s ongly sugges ha da k- and blank- i ed ke nels e lec only acciden al ma ches linking weak esidual spikes o hemodyna- mics ha a e ac ually domina ed by spike-independen ial- ela ed signals. 2.4.5 Blank sub ac ion equi ed o es ima e spikes om imaging The p ima y use o neu oimaging is as a p oxy o local neu al ac i i y. We wan ed o quan i y he impo ance o blank sub ac ion when equa ing he imaging signal wi h neu al esponse. To his end, we compa ed he alidi y wi h which we could deduce measu ed spiking om ull e sus blank- sub ac ed hemodynamics by decon ol ing (Glo e 1999) wi h he ele an op imal HRF. We expec ed ha decon ol ing a gi en blank-sub ac ed ( ha is, s imulus e oked) signal HSTIM using i s op imal ke nel HRFSTIM would i ially e u n a alid es ima e o he co esponding blank-sub ac ed ( ha is, s imulus e oked) spiking SSTIM, as hese signals we e well i ed o each o he . Wha we wan ed o assess, o compa ison, was he eliabili y wi h which he ull measu ed spiking S could be ob ained om he ull hemodyna- mics H using a simila decon olu ion wi h i s op imal ke nel HRFNULL. Gi en ha con olu ion is equi alen o he p oduc o Fou ie ans o ms in equency space, we decon ol ed by di iding he Fou ie ans o m o he hemodynamic signal by he Fou ie ans o m o he ele an HRF ke nel. As HRFs ha e e y li le powe a high empo al equencies, e lec ing he slow ime cou se o he hemodynamic esponse Si o in e al. 2009, we es ic ed 47 p oposed ea lie on empi ical g ounds, al eady do so e ec i ely; hese include designs ha linea ly sub ac ei he blank (Pes illi e al. 2011; Sh oye man e al. 2000; La sson e al. 2006) o nonspeci ic global om local signals (Donne e al. 2008; Fox e al. 2006), designs ha con as one senso y s imulus agains ano he in a common ask s uc u e (Meng e al. 2005; Cheng e al. 2001), and designs ha eg ess he hemodynamic signal agains a ange o s imulus in ensi ies (Engel e al. 1997; Boyn on e al. 1999). No ably, o such sub ac ion o p ope ly e eal s imulus- ela ed signals, he sub ac ed ial p esumably needs o be iden ical o s imula ed ials in all espec s ( iming, ewa d, e oked an icipa ion, e c.), di e ing only in no con aining he s imulus o in e es . A ano he le el, howe e , he obus link be ween s imulus-e oked hemody- namics and spiking h ows in o sha pe elie he lack o such a link o he ial- ela ed signal and sugges s ha o he non-senso y signals may be simila ly poo ly ela ed o local spiking (Jack e al. 2006; Donne e al. 2008). Fo example, he e is a well-known disc epancy be ween obus MRI e i- dence o a en ional modula ion in human V1 ( e . Ress e al. 2000) and he lack o such modula ion in elec ode eco dings om macaques (Luck e al. 1997). Ou indings aise he possibili y ha his en i e class o non-senso y signals could ha e neu al unde pinnings dis inc om senso y-d i en spiking ac i i y. The ial- ela ed signal desc ibed he e is also dis inc om he cohe en ongoing V1 ac i i y imaged in anes he ized animals (A ieli e al. 1995), as he la e was closely co ela ed wi h local spiking and LFP (A ieli e al. 1995). The ial- ela ed signal may in ol e neu omodula o y (Logo he is 2008) inpu om some b ain s em cen e ha acks beha io al iming o i may e lec eedback om some highe co ical cen e . Fu he - mo e, i could ac p e e en ially on cells o he han py amidal neu ons, such as in e neu ons o as ocy es. The e is also he possibili y o di ec neu o- modula o y con ol o blood essels gi ing ise o he hemodynamic signal (K ime e al. 1998). Finally, when in e p e ing eco ded hemodynamics as a measu e o local neu al ac i i y, i emains o be es ablished whe he he 48 s eng h o he ial- ela ed signal can be equa ed wi h ha o s imulus- e oked signals on any common measu e, such as a me abolic one o ene gy consump ion. A numbe o addi ional ques ions emain. We ha e demons a ed he MLM o a ea V1. I would be impo an o see how well i gene alizes o e o he b ain egions. Fu he mo e, we con olled spike a e by a ying he con as o la ge, uni o m g a ings, which changes ac i i y mono onically o e he popula ion o ac i a ed neu ons. We could ge aluable insigh s by con ol- ling spike a es in a manne ha di e en ially a ge s di e en neu onal popula ions, o example, by ha ing punc a e s imuli, he eby possibly ge ing di e en a es o spa ial d op-o in di e en signals (LFP, spiking, hemodynamics). We no ed ha he goodness o i be ween senso y hemo- dynamics and gamma-band LFP was much mo e a iable han wi h spiking (da a no shown), in con as wi h ea lie epo s Logo he is e al. 2001. Finding answe s o hese ques ions would be c i ical o in e p e ing unc ional b ain imaging. 2.6 Supplemen a y in o ma ion 2.6.1 Appendix: Homogeneous Linea (‘Null’) and Modi ied Linea Model (MLM) o Spike-p edic ed Hemodynamics Appendix: Homogeneous Linea (‘Null’) and Modi ied Linea Model (MLM) o Spike-p edic ed Hemodynamics Le us de ine he ull measu ed hemodynamic signal as H( ) and he ne local spiking as s( ). The homogeneous linea model (ou ‘Null’ model) p oposes ha he wo should be ela ed h ough: H = h s + n (1) whe e ( ) deno es con olu ion, h ( ) is an app op ia e hemodynamic esponse ke nel and n( ) is noise ( he explici ‘( )’ dependence is d opped in his and subsequen equa ions o a oid clu e ). By con as , he Modi ied Linea Model (‘MLM’) p oposes ha H and s a e ela ed h ough: 49 H = h s + T + n (2) whe e T is he an icipa o y ial- ela ed hemodynamic signal Si o in & Das 2009. is modeled he e as a spike-independen signal a ial pe iodici y, aligned o ial onse s and p esen in all co ec ly comple ed ials Si o in e al. 2012. We assume ha he shape o T is de e mined only by ial iming, independen o whe he he ial has a isual s imulus o a blank, o e en in ol es da k- oom ixa ion. No e ha hese equa ions a e exp essions o he unc ional o ms o he espec i e models; o any da a se , he op imal h sa is ying Eq 1 is likely o be di e en om ha sa is ying Eq 2. To es hese models agains da a we assumed ha he h ke nels – he e, ‘impulse unc ions’ quan i ying he monophasic spike- igge ed inc ease in local blood olume – can be modeled as s anda d monophasic gamma- a ia e unc ions (Cohen 1997). Op imal ke nels we e calcula ed o each model by i ing o da a h ough a s anda d leas -squa es ou ine – speci ically, i ing he app op ia e mean measu ed hemodynamic esponse, aligned o ial onse s and a e aged o e all con as s and all co ec ials, o he co esponding mean spiking esponse. (See Me hods) Fo he ‘null’ model his p ocess gene a es he ke nel de ined as ‘HRFNULL’ (bes i ing he equa ion H = HRFNULL S) whe e S is he measu ed spiking. The op imal p edic ed hemodynamics, using his model, hen becomes: (3) (A no e on nomencla u e: uppe case ‘S’, ‘HRF’ e c will be used h oughou o dis inguish he measu ed spiking and i ed ke nels om hei ‘idealized’ equi alen s (in lowe case) as in Eq 1,2. This is o o mally ecognize ha while he H in Eq 1 can be equa ed wi h measu ed hemodynamics, he measu ed mul i-uni is S only a sample o he ‘ ue’ ne local spiking ‘s’. The sample on any gi en session depends on he elec ode ip, MUA spiking h eshold e c. Wi h la e con ol expe imen s we show ha he sampling is eliable ac oss days, wi h he p ima y a iabili y being es ic ed o an 50 i ele an scale ac o . (see las sec ion o Me hods and Supplemen a y Figu e 2.11).) Fo i ing he MLM o da a i is necessa y o i s sepa a e he ial- ela ed T om spike-p edic ed hemodynamic componen s. We do so by sub ac ing blank- ial esponses om o he ials, based on ou assump ion ha T is s e eo yped and p esen in all co ec ials. In blank ials, hemodynamics HBLANK and spiking sBLANK should be ela ed by: HBLANK = h sBLANK + T + n (4) The sBLANK is he ial- ela ed blank- ial spiking no ed ea lie (Figu e 2.1b), p esumably due o uncon olled isual inpu associa ed wi h he animal’s ial-locked eye ixa ions. No e ha he MLM explici ly ela es his spiking o i s hemodynamic co ela e wi h a common h like any o he isually e oked spiking. On aking means ac oss all co ec ials, aligned o ial onse s, (deno ed by he symbol ) we ge : (5) which, on sub ac ing om Eq 2 gi es: (No e: blank- ial sub ac ion has been used empi ically ea lie in image analysis (Sh oye man e al. 2000; La sson e al. 2006)): (6) Le us de ine hese blank-sub ac ed hemodynamics and spiking as he co esponding ‘S imulus-e oked’ quan i ies HSTIM = H - HBLANK and sSTIM = s - sBLANK . Eq 6 can hen be ew i en: HSTIM = h sSTIM + n (7) The h in his model (iden ical in Eqs 2 and 4 h ough 7) can now be es ima- ed by i ing he measu ed sSTIM = S - SBLANK o he measu ed HSTIM = H - HBLANK using Eq 7, o gi e an op imal gamma- a ia e ke nel HRFSTIM. The co esponding p edic ed ‘S imulus e oked’ hemodynamics is hen: (8) This HRFSTIM can be used o ob ain he p esumed ial- ela ed signal T. We can de ine T as he mean signal ha emains a e sub ac ing, om he ull 51 measu ed H, all signal componen s p edic able om measu ed spiking whe he s imulus e oked o uncon olled (Eq 2). Wi h ou assump ion ha all isually d i en hemodynamics is ela ed o co esponding spiking h ough a common HRFSTIM, we ge : (9) Finally, o e alua e he goodness o i o each model we compa ed model p edic ions wi h hei expe imen ally ob ained co ela es. The compa isons we e quan i ied using he measu e R2, de ined: (10) whe e and a e, espec i ely, he a iance o he measu ed hemodynamics, and o he esidual e o . This la e quan i y is de ined, in gene al, as he di e ence be ween measu ed and p edic ed signals. R2 always has a alue <= 1; he close i is o 1, he smalle he esidual ela i e o he measu ed signal, and he be e he i . In cases whe e he p edic ion is made using he op imal ke nel o he gi en measu ed signal, R2 de ines he ac ion o signal a iance explained by he p edic ion, and i lies, by de ini ion, be ween 0 and 1. Fo he null model his is he case, o example, when compa ing he mean a e aged ac oss all con as s, agains he co esponding mean measu ed H since hese quan i ies a e op imally linked ia HRFNULL (see Eq 3. The esidual he e is de ined as ). Howe e , R2 is also a use ul measu e in o he compa isons, e.g. when quan i ying how well he op imal ke nel o he o e all mean signal (a e aged ac oss con as s) p edic s he signal o pa icula con as s. Fo he subse o ials a a pa icula con as he e o can be much la ge han he measu ed signal, leading o la ge nega i e R2 alues. We used he same o malism o es wo con ols ha we con as ed agains he MLM (Resul s, Sec ion 6). As he i s con ol we checked whe he he blank- ial hemodynamic signal HBLANK could be p edic ed linea ly om blank- ial spiking alone: 52 HBLANK = h sBLANK + n (11) This equa ion can be i ed using he measu ed SBLANK o o mally de ine an op imal ke nel ‘HRFBLANK’, and co esponding p edic ion . The goodness o his i could be quan i ied using a measu e R2 de ined as abo e, wi h he esidual being and he measu ed signal HBLANK. The same o malism can be used o es he con ol model ha da k- oom hemodynamics can be p edic ed using da k- oom spiking alone: HDARK = h sDARK + n (12) This equa ion can be i simila ly using he measu ed SDARK o de ine an op imal ke nel ‘HRFDARK’, co esponding p edic ion , esidual and goodness o i R2. 2.6.2 Supplemen a y Figu es Figu e 2.8 – Blank- ial spike ace SBLANK is likely isually d i en, en ained o eye ixa ion schedule. (a) Uppe panel: Same as Figu e 2.1b ( ull spiking, aligned o ial onse s and a e aged by con as ). Lowe panel: Co esponding eye acke eco ds, also a e aged by s imulus con as , showing he monkey’s eye posi ion ela i e o he ixa ion poin . No e he Time (sec) 0 2 4 6 8 10 0 25 Deg. Eye Dis . Blank (0.0) 6.25 12.50 25.00 50.00 100.00 Con as (%) Spike Ra e (Hz) 0 100 200 300 a. Spike Ra e (Hz) b. 0 20 40 60 S imulus Blank Da k Room Time (sec) Deg. 0 5 10 15 0 14 Eye Dis . SBLANK 53 s iking simila i y be ween he eye posi ion aces ( o all s imulus con as s) and he blank- ial spike ace SBLANK: high spiking be o e he animal ixa es (i.e. be o e =0 s, ial onse ); lowe spiking while he animal ixa es on he g ey moni o ( = 0 o 5.5 s) and high again when he animal looks away a he end o he ixa ion pe iod. The same spiking pa e n o ms he common baseline o all spiking esponses, independen o s imulus con as . (b) Blank- ial spiking pa e n is ex inguished in da kness, e en hough he eye ixa ion schedule is unchanged. Uppe panel: Spiking eco ds om a day wi h bo h isually s imula ed ials (one 25% con as s imulus and a blank) and da k- oom ials in he same expe imen al session. (All ials in ol ed a common pa e n o wo ixa ions, as in Figu e 2.5a-c; 9-s ixa ions, 18-s ials). Bo h he ‘s imulus’ and he ‘blank’ ials show he same common baseline o educed spiking as he animal ixa es and inc eased spiking in be ween ixa ions. No such ixa ion-linked luc ua ions in spiking can be seen in he da k- oom ials. Lowe panel: Co esponding eye-posi ion aces, essen ially iden ical ega dless o ial ype, (s imula ed o da k- oom). Pe iods o ixa ion and s imula ion indica ed on he ime line in colo key as in all o he igu es. Figu e 2.9 – Schema ic o Modi ied Linea Model (MLM). (a) S imulus- e oked spiking, shown as esponses o s imuli o di e en con as s, linea ly p edic s co esponding s imulus-e oked hemodynamics (b) Howe e , since he animal is engaged in a pe iodic ask, on e e y (co ec ) ial he e is an addi ional s e eo yped T ial-Rela ed Signal (c) ha adds on linea ly o gi e he ne hemodynamic signal (d). Impo an ly, because he T ial-Rela ed Signal adds on o e e y ial, i can also be linea ly emo ed o e eal he ‘pu e’ s imulus-e oked hemodynamics (panel b) by sub ac ing he blank- ial esponse (‘HBLANK’, panel d) om esponses a o he con as s (o he aces, panel d). The blank- ial spiking ‘SBLANK’ is shown he e as a la line in panel a. I in addi ion o he speci ic s imulus-e oked spiking he e had also been a ial- ela ed spiking esponse – p esumably due o he animal mo ing his eyes pe iodically away om and back owa ds he moni o (Figu e 2.1b, Supplemen a y Figu e 2.8) – his spiking signal (no shown 02 4 6 8 10 Time (sec) 0.01 -0.06 -0.05 -0.02 0 -0.04 -0.03 -0.01 Blank 12.50 25.00 50.00 100.00 dR/R Spike Ra e (Hz) 0 2 4 6 8 10 Time (sec) -100 0 100 200 300 S im-e oked Spiking S im-e oked Hemo 02 4 6 8 10 Time (sec) 0.04 -0.03 -0.02 0.01 0.03 -0.01 0 0.02 dR/R + T ial-Rela ed Signal =02 4 6 8 10 Time (sec) 0.02 -0.05 -0.04 -0.01 0.01 -0.03 -0.02 0 Ne Hemo Signal a. b. c. d. HBLANK SBLANK 54 he e, o a oid clu e ) would jus ha e added uni o mly o each ace in panel a. The co esponding hemodynamic esponse – p edic ed, acco ding o he MLM, using he same ke nel as he s imulus-e oked hemodynamics – would ha e added uni o mly o each ace in panels b and d including, in pa icula , HBLANK. On sub ac ing he ne blank- ial esponse HBLANK hese o he ial- ela ed signals would also ge linea ly sub ac ed away e ealing jus he s imulus-e oked signals (Figu e 2.2a,b). T ial iming is indica ed as usual in all panels, wi h pe iodic ixa ion, s imulus p esen a ion and end o ial (i.e. s a o nex ial). S imula ion pe iod is no indica ed in panel c o emphasize ha he ial- ela ed signal is d i en by he ask s uc u e alone. 55 a. HRFST IM : Modi ied Linea Model HRFNULL : Null Model e. -1 -0.5 0 0.5 1 1 0 0.5 -0.5 -1 Null Model Modi ied Linea Model (MLM) R2 mean d.1 0 0.5 -0.5 -1 -1.5 -2.5 -2 -3-1 -0.5 0 0.5 1 Null Model Modi ied Linea Model (MLM) A e age o R2 pe con as Con as : 6.25% 60 0 10 20 30 40 50 0 0.2 0.4 0.6 0.8 1-1.2 -1.0 -0.8 -0.6 -0.4 -0.2-1.4 b. A e age o pe con as R2 0 0.2 0.4 0.6 0.8 1 R2 15 30 45 60 75 90 0 c. 200 0 0.2 0.4 0.6 0.8 1 R2 40 80 120 160 0 R 2 mean R2 0 0.2 0.4 0.6 0.8 1 0 50 40 30 20 10 Con as : 12.5% 0 0.2 0.4 0.6 0.8 1 0 40 30 20 10 Con as : 25% 0 0.2 0.4 0.6 0.8 1 0 80 60 40 20 Con as : 50% 0 0.2 0.4 0.6 0.8 1 0 120 90 60 30 Con as : 100% Modi ied Linea Model Null Model O iginal Median 95% con dc O iginal Median 95% con dc 6.25% 0.94 0.89 [0.73; 0.96] -0.12 -0.23 [-0.81; 0.18] 12.5% 0.94 0.91 [0.73; 0.97] 0.26 0.26 [0.06; 0.41] 25% 0.87 0.88 [0.74; 0.96] 0.46 0.47 [0.37; 0.56] 50% 0.96 0.94 [0.87; 0.98] 0.58 0.57 [0.48; 0.66] 100% 0.95 0.94 [0.90; 0.97] 0.59 0.61 [0.55; 0.66] A g Con 0.93 0.91 [0.86; 0.94] 0.35 0.34 [0.22; 0.43] Mean 0.99 0.99 [0.86; 0.94] 0.49 0.49 [0.43; 0.54] R2 56 Figu e 2.10 – Compa ing con idence in e als o he goodness o i R2, o he Modi ied Linea Model (MLM) s. he Null Model, boo s ap esul s. (Table and a-c: o same example da a se as in Figu e 2.1, Figu e 2.2; d,e: popula ion). (a) Supe imposed his og ams o R2 dis ibu ions, o he Modi ied Linea Model (MLM), in ed, and he Null Model in o ange (200 boo s ap uns, using se s o ials chosen andomly wi h eplacemen , each se p ocessed using bo h MLM, i.e. i ing HRFSTIM, and Null Model, i.e. i ing HRFNULL, and calcula ing co esponding R2. See Me hods o de ails). R2 calcula ed sepa a ely pe con as , as in he alues labeling indi idual aces in Figu e 2.2c. No e ha while he his og ams o he di e en con as s ha e he same X scale, he 6.25% con as panel has di e en X limi s because a his con as he R2s ex end in o nega i e alues. N = 261 ials o al, oughly 43 pe con as . Table summa izing he boo s apping esul s o he example da a se . O iginal e e s o he R2 ob ained o he ull (o iginal) da ase . Median designa es he median o he boo s apping p ocedu e, and ‘95% con dc’ he 95% con idence in e al calcula ed o he boo s apping. Numbe s in le -mos column e e o s imulus con as s; ‘A g Con ’ e e s o he a e age ac oss all con as , i.e. as in panel b, and ‘Mean’ e e s o he mean R2 i.e. as in panel c. (b) R2 alues om panel a, a e aged ac oss con as s. Same calcula ion as in Figu e 2.2d. (c) Mean R2 o he same boo s ap ial selec ions as in panels a,b bu calcula ed a e i s a e aging he ials ac oss con as s (calcula ions as in Figu e 2.2c inse ). (d) Compa ing boo s ap-de i ed es ima es o R2 o Null s. MLM o e popula ion o expe imen s. Calcula ed sepa a ely by con as and hen a e aged, as in panel b. We used 80% con idence limi s in he popula ion da a, o accommoda e some da a se s wi h long ails in boo s ap dis ibu ions. Da a poin s show he medians o boo s ap es ima es while e o ba s show co esponding 80% con idence limi s. Same da a as shown in Figu e 2.2e. No e, howe e , ha he da a poin s in Figu e 2.2e indica e R2 o he expe imen ally ob ained se o ials and a e hus consis en ly sligh ly highe han he medians o he co esponding boo s ap es ima es he e. (e) Same, bu o ‘Mean R2’ calcula ed as in panel c. Same da a as shown in Figu e 2.2d (again, da a poin s sligh ly di e en om Figu e 2.2d since Figu e 2.2d shows R2 o ac ual ial se ). 63 The da a p esen ed he e concu s wi h he iew ha he hemodynamic esponse in eg a es ac i i y om a a ie y o sou ces. Conside ing local neu al ac i i y as he sole p edic o o hemodynamic ac i i y does no gi e an accu a e pic u e o he esul ing esponse. 3.2 Me hods Me hods used he e we e he same as desc ibed in Si o in & Das 2009. Simul aneous in insic signal op ical imaging and elec ophysiology we e acqui ed om ale macaques engaged in di e en asks. All expe imen s p esen ed he e we e pe o med in da kness o quasi- o al da kness. All expe imen al p ocedu es we e pe o med in acco dance wi h he US Na ional Ins i u es o Heal h Guide o he Ca e and Use o Labo a o y Animals and we e app o ed by he Ins i u ional Animal Ca e and Use Commi ees o Columbia Uni e si y and he New Yo k S a e Psychia ic Ins i u e. De ailed me hods desc ip ion o he expe imen al me hodology used in he lab can be ound elsewhe e (e. g. Si o in & Das 2009, Ca doso e al. 2012, Lima e al. 2014). He e a e only e e enced he me hods ha a e pa icula ly ele an o he da a p esen ed below. The da a included in his chap e e lec s a combina ion o di e en expe imen s, hence speci ic expe imen al de ails o analysis a e included when app op ia e h oughou he ex . Imaging da a was acqui ed a ei he 7.5 o 15 Hz, a his a e pulsa ion is obse able in he imaging esponse. Co ical pulsa ions we e emo ed by low-pass il e ing, using he Ch onux MATLAB Toolbox unc ion unline.m ( ypical hea a es we e ~2–3 Hz, much as e han he ypical hemodyna- mic esponse equencies o ~<0.5 Hz). Hea a e was es ima ed om he pulsa ion. Each pulsa ion e en has a peak and ough; he dis ance be ween peaks and oughs was used o iden i y law ul pulsa ion e en s. Hea a e was es ima ed as he equency o he pulsa ion e en s. Slow empo al d i s (>30 s) we e emo ed wi h high-pass il e ing (using he 64 Ch onux MATLAB Toolbox unc ion unline.m), excep : when calcula ing he powe spec um o he hemodynamic esponse on he a iable in e - ial- in e al manipula ion and when looking a he slow changes on he hemodynamic esponse as he animals p og ess in o blocks o high o low ewa d, in he expe imen s o manipula ion o ewa d amoun . In hese excep ions, a linea end ( eg ession on he ull ime se ies) was sub ac ed om he esponse. Excep ions desc ibed in he ex . In he sec ion whe e we manipula ed ixa ion du a ion, one o he ixa ion du a ion schedules ha we p esen ed obeyed he ule o a ‘memo yless’ p ocess. By memo yless p ocess i is mean ha he p obabili y o an e en happening in he u u e gi en i has no ye happened is independen o he ime ha has passed since he beginning o ha p ocess. Mo e o mally: p(x > a + b | x> a) = p(x> b). Using he ules o condi ional p obabili ies: p(x > a + b | x> a) = p(x> a + b  x> a) / p(x> a), and as ime is con inuous: p(x> a + b  x> a) = p(x> a + b), he e o e i comes ha p(x > a + b | x> a) = p(x> b) = p(x> a + b)/ p(x> a), hus p(x> a + b) = p(x > a) * p(x > b). No demons a ed he e, bu ma hema ically his uni e sally will be alid i he p(x> i) will be gi en by an exponen ial unc ion: p(x> i) = ( he p obabili y will be be ween 0-1, we will use a andom numbe be ween 0-1 o gene a e samples o , ixa ion du a ion). We sampled ixa ion du a ions wi hin ixed ime in e als ( he e was a minimum ixa ion ime, min and a maximum one, max) = min –  *log( and[0;1]) and  max (i > max, hen a new is calcula ed, un il  max). 3.3 Va iable T ial Du a ions Aiming a cha ac e izing di e en aspec s o he ask- ela ed esponse, we p esen da a collec ed simila ly o da a in Si o in & Das 2009, wi h small a ia ions men ioned below. Subjec s pe o med a ixa ion ask in quasi o al da kness wi h he moni o ha ing only a small ixa ion poin exposed 65 (o he wise all ligh om moni o was blocked, as in he o iginal s udy by Si o in & Das 2009). In his manipula ion o he ask, he subjec was asked o pe o m he ixa ion ask in wo di e en schedules. Ei he he schedule was o sho ials, wi h a sho e in e - ial in e al, ITI, ( om 6 o10 s, only one sho ITI used in each session), o long ials, wi h a longe ITI (12 o 20 s, only one long ITI used in each session). The ask he animal pe o med was a ixa ion ask. The colo o he ixa ion poin cues he subjec o when o aim o ixa ion. Once he animal achie es ixa ion he colo o he ixa ion poin changes and once he ixa ion pe iod ends he ixa ion poin u ns o and he animal ecei es a liquid ewa d o holding ixa ion. Finally, he e is a ime delay (ITI) be ween ials. T ials in which he animal did no hold ixa ion o he eques ed pe iod we e no ewa ded. T ials we e p esen ed in di e en schedules, whe e ITI was changed. The di e en schedules we e p esen ed in blocks. In gene al, we eques ed he subjec s o pe o m 20 comple e ials un il he schedule changed (s a ing a new block). The animal was no cued o he change in ITI schedule. Ses- sions could s a wi h a sho o long ITI ( he o de o he i s ITI, i long o sho , did no a ec he esul s). Once again, he subjec ’s ask was o hold ixa ion o he pe iod he ixa ion poin had a pa icula colo ( he subjec was eques ed o hold ixa ion o 2-4 s, a ying be ween sessions). We eco ded he hemodynamic esponse using in insic signal op ical imaging, al eady desc ibed, and simul aneous elec ical eco dings in he imaging egion. Fu he mo e, om he imaging esponse we ex ac ed he hea a e esponse (see me hods). Based on wha we know abou he ask- ela ed esponse om p io wo k done in ou labo a o y (Si o in & Das 2009), we hypo hesized ha , wi hin he o al ial leng hs ha we ha e used he e, he ask- ela ed esponse should, on a e age, en ain o he ial du a ion. S ill se e al aspec s o his en ainmen a e no unde s ood. One o he ques ions we had abou he ask- ela ed esponse was: wha a e he e en s ha igge he en ainmen o he hemodynamic esponse? 66 Wi h his sys ema ic al e na ion o blocks o sho and long ials, we wan ed o e alua e i he e we e sys ema ic changes in he ime cou se o he esponses. In pa icula we wan ed o know i he e a e changes ea ly in a sho s. a long ial, up un il he du a ion o he sho e ITI. Beha io ally, ials should be simila (un il he du a ion o he sho e ITI) as he ask he animal has o do is he same: ixa ion. Up un il when he du a ion o he ITI does no exceed ha on a sho ITI ial, all he ials (sho o long ITI) a e indis inguishable. The aim was o unde s and wha aspec s o ials (e.g., e o o ixa ion, ewa d) could in luence he hemodynamic en ainmen , i.e., wha beha io al e en s could be in ol ed in elici ing o igge ing he hemodynamic ask- ela ed esponse. Simul aneous o he eco ding o he hemodynamic esponse, we eco ded MUA ac i i y o one o wo elec odes in he same egion whe e we eco ded he neu oimaging da a. We also ecoded he hea a e esponse; econs- uc ed om he b ain pulsa ion obse ed in he neu oimaging esponse. We looked o concomi an changes in ei he o hese addi ional me ics wi h he hemodynamic esponse. Figu e 3.1 shows a ypical session (11 sessions we e eco ded o one sub- jec , monkey T). In Figu e 3.1a is he mean hemodynamic esponse o sho ( op) o long (middle) ials. I can be no iced ha he esponses o sho o long ials a e di e en , bu he p o iles up un il he du a ion o he sho ial a e simila (Figu e 3.1a bo om panel). Fo he session shown, he mean ial du a ion o a sho ial was 15.92  0.02 s and o a long ial was 23.95  0.03 s (changes in ial du a ions e lec small changes in he ime he animal akes o achie e ixa ion on each ial). Looking a he spec al p o ile o he hemodynamic signal (Figu e 3.1d) i can be no iced ha he e a e peaks in his spec um ha closely ma ch he wo schedules o comple ed ials (15.2 and 21.4 s, o sho and long ials, espec i ely). The e a e o he peaks in he spec um no di ec ly ela ed o he schedule o comple ed ials, hough. I should be men ioned ha incomple e ials had sho e du a ions because he subjec did no hold 67 ixa ion du ing all he eques ed pe iod and no ewa d is adminis e ed, These ime windows we e no discoun ed om he du a ion o he in e - ial in e al. Incomple e ials ha e less s e eo ypical du a ions as he animals in e up ials a di e en momen s. Fo all sessions he e we e mo e comple e han incomple e ials. S ill, he p esence o peaks a highe equencies han ei he sho o long ial equencies could be ela ed o incomple e ials. The p o ile o he hea a e esponse in Figu e 3.1b was also simila o he wo ial schedules, and di e en om he hemodynamic esponse. I is wo h no ing ha he hea a e spec um o his session did no show clea peaks a he di e en ial schedule equencies (Figu e 3.1e). Finally, he spiking a es o a mul i-uni elec ode loca ed he same egion as he imaging did no show a signi ican pa e n associa ed wi h his ask (Figu e 3.1c), as expec ed om Si o in & Das 2009. Fo his session, he MUA spec um had mul iple peaks, bu no igh ly ma ching he ial schedules (Figu e 3.1 ). 68 Figu e 3.1 – Single session showing ha di e en ial schedules ha e associa ed di e en hemodynamic, hea a e and MUA esponses. (a) Hemodynamic esponses o sho ials ( op), long ials (middle) and o e lay o a e age sho ials (black) and long ials (g ay) (bo om). E o 0 0.1 0.2 -0.2 0 0.2 0.4 0.6 F equency (Hz) Log Powe Spec um 8.5 5.85.1 11.3 18.1 30.8 0 0.05 0.1 0.15 0.2 -8 -6 -4 -2 F equency (Hz) Log Powe Spec um 14.7 11.9 8.0 0 0.05 0.1 0.15 0.2 -12 -10 -8 -6 F equency (Hz) Log Powe Spec um 21.4 15.2 12.5 010 20 0 50 100 150 Spiking (Hz) ime (sec) ela i e o ial s a 010 20 1.4 1.6 1.8 2 Hea a e (Hz) ime (sec) ela i e o ial s a 010 20 -0.01 -0.005 0 0.005 0.01 -dR/R ime (sec) ela i e o ial s a 010 20 0 50 100 150 Spiking (Hz) 010 20 1.4 1.6 1.8 2 Hea a e (Hz) 010 20 -0.01 -0.005 0 0.005 0.01 -dR/R 010 20 0 50 100 150 Spiking (Hz) 010 20 1.4 1.6 1.8 2 Hea a e (Hz) 010 20 -0.01 -0.005 0 0.005 0.01 -dR/R ed cba 69 ba s calcula ed as s.e.m.. The ed ho izon al lines indica e he pe iod he animal was eques ed o hold ixa ion. (b) Simila o a, bu o he hea a e esponse. (c) Simila o a, bu o he mul i-uni spiking ac i i y. (d) Log- powe spec um o he hemodynamic esponse. (Powe spec a we e calcula ed on he da a wi hou discoun ing o slow d i s ha migh occu h oughou he imaging session, ha we e in gene al always used, unless o he wise s a ed, as he e.) The ed c osses indica e peaks a ound he ials’ pe iodici ies; he numbe s indica e he co esponding pe iod. (e) Log-powe spec um o he hea a e; simila o d. ( ) Log-powe spec um o he MUA ac i i y, simila do d. F om obse ing he single case p esen ed in Figu e 3.1 i is no iceable he appa en simila i y o he hemodynamic esponse du ing some ini ial pe iod in which sho and long ials a e indis inguishable. We he e o e wen on o quan i y he simila i y in he shape o he hemodynamic p o ile wi hin he i s seconds o comple ed sho o long ials. We looked a he ini ial 10 s o all comple ed ials, ei he o ials coming om blocks o sho o long ials (n = 11 sessions). We selec ed his in e al because i was smalle han he leng h o a sho ial ( o all conside ed sessions). Fo his quan i ica ion we i s compu ed he Pea son’s co ela ion coe icien o he ini ial 10 s o all pai s o sho ials; and he Pea son’s co ela ion coe icien o he ini ial 10 s o all pai s o long ials. We es ablished he dis ibu ion o co ela ion coe icien p o iles o sho -sho and long-long ials. We hen calcula ed he Pea son’s co ela ion coe icien o he ini ial 10 s o all he pai s o sho and long ials. We used he p e ious co ela ion coe icien s o he pai s o sho -sho and long-long ials o ha e a compa ison baseline o he co ela ion coe icien s o sho -long ials. Ou null hypo hesis was ha he dis ibu ion o co ela ion coe icien s o sho -long ials is indis inguishable om he dis ibu ion o sho -sho (o long-long) ials. In Figu e 3.2a, i can be no iced ha he dis ibu ion o Pea son’s co ela ion coe icien s o sho - sho ials o long-long ials a e simila . In his session he hemodynamic ask- ela ed esponse is ai ly s e eo ypical: he co ela ion p o ile is skewed owa ds 1. Mo e impo an ly his dis ibu ion was also simila o compa i- 70 sons be ween sho -long ials (ligh e g ay, Figu e 3.2a). Independen ly o he pa icula shapes o he ask- ela ed esponse on each session, o how s e eo ypical his esponse was wi hin a session, we obse ed ha he dis ibu ion o ou simila i y measu e ( he Pea son’s co ela ion) was p ese ed (wi hin a session). The simila i y be ween sho and long ials was independen o he a iabili y o he ask- ela ed esponse wi hin each session. Fo each session we es ed how simila he dis ibu ions we e o sho -sho ials e sus long-long ials (Figu e 3.2b); no session eached signi icance. Then we used a non-pa ame ic Wilcoxon signed- ank es o compa e he dis ibu ion p o iles o co ela ion coe icien s sho -sho s. long-long s. sho -long ials. Once again he compa ison o he shapes o he dis ibu ion o co ela ion coe icien s o sho -long ials agains ei he sho -sho o long-long ials did no each signi icance o any session (Figu e 3.2c). This indica es ha he h ee dis ibu ions we e no s a is ically di e en . I suppo s he hypo hesis ha o he ini ial du a ion o sho o long ials he hemodynamic esponses a e indis inguishable (wi hin he ial schedules ha we ha e es ed). Figu e 3.2 – The ini ial ew seconds (10 s) a e indis inguishable o sho o long ials. (a) Fo he same session as in Figu e 3.1, he compu ed co ela- ion be ween he ini ial 10 s o all comple ed ials, ei he long o sho . In black is he co ela ion p o ile be ween he ini ial seconds o sho ials ( ials sha e a signi ican simila i y, he e o e co ela ion ends o be high: close o 1). In da k g ay is he co ela ion p o ile be ween he ini ial seconds a b c 0 0.5 1 0 1 2 3 p- alue numbe o sessions 0 0.5 1 p- alue numbe o sessions 0 2 4 -1 0 1 Co ela ion be ween ial pai s Po ion o ials Sho -Sho Long-Long Sho -Long 0 0.1 0.2 71 be ween long ials. In ligh g ay is he co ela ion p o ile be ween he i s ew seconds o long and sho ials. Each co ela ion p o ile was no malized by he o al numbe o pai s compa ed. (b) Popula ion da a summa y o 11 sessions. p- alue o he compa ison be ween he co ela ion shapes o sho o long ials. No e: ed line ma kes signi icance le el (signi icance alue, , co ec ed o mul iple compa isons; co ela ion p o iles we e binned in 21 in e als: Bon e oni co ec ion, esul ing  = 0.05/21); no session is below he signi icance le el. (c) Simila o b, bu o he combina ion be ween he co ela ion shapes o sho -long ials ela i e o ei he sho o long ial only co ela ion dis ibu ion ( o al o p- alues: 22, wo o each o he 11 sessions). No e: no session eaches signi icance (same signi icance le el as in b). We obse ed ha he hemodynamic esponse ended o en ain o he ial schedule, bu he ini ial esponse o sho o long ials (which con ains he ini ial inc ease in blood olume) was indis inguishable o he wo condi ions. We canno dis ega d ha o di e en ial schedules (much sho e , much longe o la ge di e ence be ween sho and long ials) he hemodynamic esponse o sho and long ials could be dissimila . Las ly, o hese se o expe imen s we wan ed o unde s and i some o he obse ed peaks in he powe spec um could be ela ed o he pe iodici y o sho and long ials. Fu he mo e, we wan ed o know i ei he o he ac i i y measu es (hemodynamic, hea a e o MUA) had spec al peaks ha be e ma ched he ial schedule, han he o he me ics. In his a emp we gene ously looked o 20 peaks in he powe spec um, below 0.5 Hz. The numbe o peaks ound pe me ic (hemodynamic, hea a e o MUA) pe session was a iable. Some sessions did no p esen clea peaks below 0.5 Hz ( his is he eason o he di e en numbe o da a poin s in Figu e 3.3b- d). Nex , o each session, we compu ed he minimal dis ance be ween he spec al peaks and he equencies co esponding o sho and long ials (Figu e 3.3a, in seconds). In his sample o sessions he e was no clea supe io i y o he hemodynamic esponse in ha ing peaks ha ma ched he sho and long ITIs. This me ic is a he c ude; he spec a did no always 72 ha e clea peaks. I sugges s ha spec a migh be noisy o he en ainmen o he di e en measu es wi h ial pe iodici y, e en hough weak, happens o all measu es ( o hese ial du a ions). In Figu e 3.3b-d we plo ed he peak posi ions agains he pe iod o he ials (sho o long). The peak posi ions we e chosen o be he closes o he ac ual ials du a ions (sho o long). The di e en panels (In Figu e 3.3b-d ) e lec he di e en me ics, hemodynamics, hea a e and MUA. I he spec al peaks would ully e lec he ial du a ion, hen he dis ance be ween he closes spec al peak and he ial pe iod should be ze o; he e is howe e la ge dispe sion. The peaks’ posi ions end somewha o ollow one ano he on he di e en me ics. Hea a e had he lowes numbe o de ec able peaks (below 0.5 Hz) and i s ela ionship wi h he o he me ics was he leas uni o m. Hea a e seems o en ain less wi h ial du a ion, when compa ed wi h he hemodynamic esponse o mul i-uni ac i i y. Figu e 3.3 – Powe spec um peaks only pa ially e lec he di e ence in ial leng hs. (a) His og am o dis ance (in seconds) be ween he leng hs o b c d 10 15 20 25 30 0 10 20 30 HEMO HR 010 20 30 0 10 20 30 HR SPKS 10 15 20 25 30 HEMO 0 10 20 30 SPKS -2 0 2 4 6 8 10 12 14 16 0 10 20 a ime dis ance (s) coun s SPKS HEMO HR 79 in e als, ha explains why he e a e mo e segmen s han expe imen al sessions). In black a e he segmen s ha had a longe peak o longe ixa ions, in ed he sessions whe e he hemodynamic esponse o he longe ixa ion peaked be o e he one o he sho e ixa ion. No e ha mos lines a e blue. Do ed segmen s we e used o monkey T, solid ones o monkey S and do ed-in e up ed ones o monkey E. (b) The dis ibu ion o he ampli udes associa ed wi h he peaks in a. The peak ampli udes we e z- sco ed so hey could be ep esen ed on he same plo . Same colo code and line s yle as in a. (c) Simila o a, bu o he hea a e peak esponse (he e we conside ed he maximum peak, independen ly o how many peaks he a e age esponse exhibi ed). (d) Simila o b, bu o he hea a e peak ampli ude. (e) Sca e plo o he peak posi ions o he hemodynamic e sus hea a e esponses showing, o each expe imen al session, he alue o he longe ixa ion ela i e o ha o he sho e (se o ze o). Lines join sho -long pai s o each session. No e: esponses end o be p edominan ly in he i s quad an . Dashed, solid and do ed-in e up ed segmen s used o monkeys T, S and E, espec i ely. ( ) Simila ly o wha was done o he peak posi ions in e, he e o he peak ampli udes. The da a p esen ed comes om se e al a ia ions on he ixa ion ask, wi h only a small numbe o sessions o each manipula ion. Ne e heless i opens in e es ing ques ions pe aining he ask- ela ed esponse: could he e o o ixa ion o an icipa ion o ewa d, o example, play a ole in igge ing his hemodynamic ask- ela ed esponse? 3.5 Va iable Rewa d Amoun P e iously we manipula ed aspec s o he iming o he ials; bu wan ed o u he explo e beha io al co ela es o he mean hemodynamic ask- es- ponse. We se up a sligh ly di e en ask wi h he goal o al e ing he ewa d le el ha he animals expe ienced, and hus explo e po en ial hemodynamic ask- ela ed esponse changes. We manipula ed he ewa d amoun ha he subjec ecei ed in each ial, in a block wise manne (no addi ional cues). O he wise he animals pe o med he same ixa ion ask. 80 As in he p e ious manipula ions, he ixa ion ask was p esen ed in quasi- o al da kness. Rewa d was manipula ed in blocks; blocks o ei he o low o high ewa d. The e was a leas a wo old di e ence be ween he ewa d le els. The block s uc u e we imposed equi ed he animal o comple e 10 ials be o e he ewa d le el changed. Beha io ally, he animals ended o be sensi i e o he le el o ewa d; b eaking ixa ion o engaging in ials less equen ly when he ewa d was lowe , as shown in Figu e 3.6. In Figu e 3.6, he p obabili y o comple ing ials o low ewa d, in ed, was in gene al lowe han o high ewa d ials, in blue. I should be men ioned ha he ideal beha io is o comple e ials independen o ewa d le el. Incomple e ials we e no ewa ded and incu ed in a ‘punishing’ in e - ial-in e al un il he s a o a new ial (no ac i e punishmen s o addi ional ime de ails we e included). In 3 o he 35 sessions he pe o mance in blocks o low ewa d ials was highe han o high ewa d ones. Figu e 3.6 – Beha io o he ixa ion ask in blocks o high and low ewa d. In ed is he ac ion o comple ed low ewa d ials and in blue he ac ion o comple ed high ewa d ials. Each pai o lines ( ed-blue) is o one session. (3 monkeys: S: 6 sessions, T: 19 sessions and E: 10 sessions.) Once mo e we looked a ial a e age hemodynamic ask- ela ed esponse. In Figu e 3.7a-d a e illus a ed esponses o a single session ( o monkey T). In Figu e 3.7a-c a e he a e age ial igge ed hemodynamic esponses o Monkey S Monkey T Monkey E 0 0.5 1 Pe o mance 81 low and high ewa d ials. I could be no iced ha o his session he magni ude o he hemodynamic esponse o high ewa d ials was highe han o low ewa d ones. This was no accompanied by concu en spiking changes (Figu e 3.7d). We wan ed o know i he same end was obse ed o he popula ion; we es ed he magni ude o high and low ewa d ials. To es o signi icance, we used a Wilcoxon signed- ank es (p- alue=6.7x10-6) indica ing he magni ude di e ence be ween high and low ewa d o be di ec ional (Figu e 3.7e). Magni ude o low ewa d ials was on a e age smalle han o high ewa d ones. a b c d e 0 0.005 0.01 0.015 0 0.005 0.01 0.015 S d o LOW ewa d ials a e discou ing DC con ibu ion S d o HIGH ewa d ials a e discou ing DC con ibu ion Y = 0.96 X - 0.47x10 R = 0.927 -3 2 Time (s) ela i e o ixa ion onse 0 5 10 15 -0.02 0 0.02 Low Rewa d -dR/R 0 5 10 15 -0.02 0 0.02 Time (s) ela i e o ixa ion onse -dR/R Hi Rewa d -8 -4 0 4 8x 10-3 -dR/R 0 5 10 15 Time (s) ela i e o ixa ion onse 0 5 10 15 -20 0 10 20 30 40 Time (s) ela i e o ixa ion onse Fi ing a e ( ela i e o mean) -10 Low High 0 0.01 0.02 0.03 -dR/R mean ial ampli ude Rewa d 82 Figu e 3.7 – Magni ude o he hemodynamic ask- ela ed esponse changes wi h ewa d amoun . (a) Hemodynamic ask- ela ed esponse o an indi i- dual session. In g ay he indi idual esponse on each comple ed ial o low ewa d; in ed is he a e age esponse. (b) Simila o a, bu conside ing all comple e ials o high ewa d in he same session; in blue is he a e age esponse. (c) O e lay o he a e age esponses in a and b. E o ba s ep esen s anda d e o o he mean. (d) Simila o c, bu o he elec o- physiological da a. No e ha he di e ence be ween he wo dis ibu ions o spiking esponses is only signi ican ly di e en a one ime poin , in his session, which p obably coincides wi h ewa d deli e y (no consis en ly obse ed o o he sessions). (e) Hemodynamic ampli ude (peak o ough) o low and high ewa d ials. Each line ep esen s one session, whe e he peak o ough ampli ude o he a e age low ewa d ial is on he le and he co esponding peak o ough o high ewa d is on he igh , o he same session. In ed a e he sessions whe e he epo ed ampli ude o low ewa d ials was highe han o high ewa d ones. Signi icance es : Wilcoxon signed ank ask, p- alue: 6.7x10-6. ( ) Noise a iabili y measu e be ween low and high ewa d ials. Sca e plo o he mean s anda d de ia ion (in ime) o low e sus high ewa d comple ed ials (be o e es ima ing he s anda d de ia ion o each ial, o subsequen a e age, o each ial he mean esponse on he ial was sub ac ed, as in panels a-c). The 90% con idence in e als we e es ima ed om boo s apping (100 epe i ions). Red line ep esen s a eg ession line h ough he da a (pa ame e s indica ed on he plo ’s inse ). No e ha his eg ession is below he equali y line. We looked a he ime dynamics o he ask- ela ed signal in low and high ewa d ials. We no iced ha he ials in lowe ewa d blocks had highe a iabili y in ime, han in highe ewa d blocks, (Figu e 3.7 , a eg ession line h ough he da a is below he equali y line). Taken oge he , we obse ed changes in he hemodynamic esponse wi h ewa d le el. Bo h he esponse magni ude and i s a iabili y in ime changed wi h ewa d le el: highe esponse magni ude and lowe ime a iabili y o high ewa d ials and lowe esponse magni ude and highe ime a iabili y o low ewa d ials. 83 Finally, we no iced ha be ween blocks o low and high ewa d ials, he e we e slowe luc ua ions in he hemodynamic signal. To quan i y his obse a ion we ook he ini ial hemodynamic signal change and emo ed only a linea end om his signal and no o he low equency changes. Fo each ial we calcula ed he mean hemodynamic esponse o each ial (ac oss he du a ion o each ial). As exempli ied o he single session in Figu e 3.8a,c. i shows a segmen o a session; one could no ice ha mean hemodynamic esponse luc ua es wi h ewa d amoun as one mo es along a block o low o high ewa d. In his da ase , as one goes along a low ewa d block he mean hemodynamic le el ended o inc ease (inc eased blood olume le el changes), and as one mo ed along a high ewa d block, he mean hemodynamic le el ended o dec ease (dec ease in blood olume changes). Looking in mo e de ail a wha was happening a he block le el, as we illus a e in Figu e 3.8c, whe e we looked a he same segmen o da a as in Figu e 3.8a, bu whe e only he mean ial esponse o comple ed ials is ep esen ed. We no iced ha along a block he mean ial hemodynamic esponse o each ial seemed o e ol e in a mono onic way. To e alua e i mean ial esponses changed sys ema ically wi h p og ession along a ewa d block, a eg ession line was i o he mean ial esponse o he i s 10 comple ed ials in each block ( ewa d blocks wi h ewe comple ed ials we e no included, and o he ones wi h highe numbe o comple ed ials, only he i s 10 comple ed ials we e conside ed). The eg ession lines a e illus a ed by he black lines on he plo Figu e 3.8c. To es i he e we e sys ema ic di e ences be ween eg ession lines o each o he ewa d le els (low o high), he slopes o he eg ession lines o all sessions we e combined, seg ega ed by ewa d le el on each block, and we es ed i he wo dis ibu ions (slopes o low ewa d blocks and slopes o high ewa d blocks) we e he same, using a Wilcoxon ank-sum es . In Figu e 3.8d, i can be no ed ha he dis ibu ions a e di e en , indica ing ha along a block o lowe ewa d he hemodynamic mean ial esponse ended o inc ease and con e sely o high ewa d blocks. 84 Figu e 3.8 – Rewa d modula es no only he a e age hemodynamic espon- se as well as he mean hemodynamic le el. (a) Hemodynamic esponse on a segmen o a session (same session a in Figu e 3.7a-d). The do ed lines ep esen he sepa a ions be ween blocks o low and high ewa d, ewa d le el on each block is indica ed by he ho izon al colo ed lines on he bo om; ed o low ewa d blocks, and blue o high ewa d ones. (b) Dis i- bu ion o he ampli ude o he mean hemodynamic esponse o each comple e ial (no malized by he s anda d de ia ion o he hemodynamic esponse change o he same session as in a), sepa a ed by ewa d amoun (n = 7462 and 7315, low and high ewa d ials) (Addi ionally, in 29 o he 35 session he blood olume was lowe on he a e age o ials o high ewa d ials han o low ewa d ones.) Wilcoxon ank-sum es on he wo dis ibu ions: p- alue = 4x10-89 < 0.05/(7462x7315), using he conse a- i e co ec ion o di iding he a e signi icance alue by o he o al possible combina ion o compa isons. (c) Same segmen as in a, he e only comple e a b c d 2000 3000 4000 -0.02 0 0.02 Time (s) -dR/R 1000 -4 -2 0 2 4 6 x10-3 Low High Rewa d Slopes o mean signal (-dR/R) along a block 4000 -0.02 Time (s) 3000 2000 1000 -0.01 0 0.01 -dR/R -2 0 2 4 Low High Rewa d no malized DC ampli ude (-dR/R) 85 ials a e conside ed; he hemodynamic esponse o each ial is ep esen- ed by he mean esponse on ha ial (in ed low ewa d ials, in blue o high ewa d ones). Do ed e ical lines (as in a) show ansi ions be ween ewa d blocks. Solid black lines ep esen indi idual eg ession lines ha we e i ed o each indi idual ewa d block (only blocks wi h 10 o mo e comple e ials we e included). No e: p edominan ly posi i e slopes in blocks o low ewa d ials, e sus he p edominan ly nega i e ones o high ewa d blocks. (d) Dis ibu ion o he slopes o all he conside ed blocks om all sessions seg ega ed by ewa d le el on he block (n = 35 sessions, 487 blocks o low ewa d and 565 blocks o high ewa d). The dis ibu ions es ed wi h Wilcoxon ank-sum es : in p- alue = 5x10-63 < 0.05/(487x565), using he conse a i e co ec ion o di iding he a e signi icance alue by o al possible combina ions o compa isons. This manipula ion o ewa d amoun clea ly showed ha he hemodynamic esponse in isual co ex changes wi h ewa d le el. The hemodynamic ask- ela ed esponse had highe magni ude and lowe a iabili y o high ewa d ials han low ewa d ones. These e ec s ode on op o he slowe changes obse ed on he hemodynamic esponse, speci ically, he mean hemodynamic signal ended o luc ua e slowly (we measu ed i a he ial pe iod) and dec eased as he subjec p og essed in o a high ewa d block, and hen inc eased as he subjec p og essed in o low ewa d blocks. The e is ex ensi e esea ch on he e ec s o ewa d in he b ain. I has been shown in BOLD MRI expe imen s wi h human subjec s ha inc eased ewa d p obabili y is associa ed wi h inc eased s imulus BOLD MRI ac i i y (Se ences 2008). Con as ing wi h hese and ou esul s, he e is wo k using con as -agen -enhanced MRI echniques as well as D1-selec i e dopamine an agonis in ale non-human p ima es pe o ming a ange o isual asks (A senaul e al. 2013). This wo k was ocused on changes in isual a eas as he animals pe o med isual asks. Thei obse a ions con adic some- wha we p esen he e; hey obse ed a selec i e dec ease in MRI ac i i y when compa ing ewa ded o un- ewa ded condi ions, and hey we e able o ela e hese changes o D1 dopamine ecep o ac i i y (using a D1-selec i e an agonis hey go smalle e ec s). The e is a signi ican expe imen al 86 di e ence be ween A senaul e al. 2013 expe imen s and Se ences 2008 (and ou s), hey bo h used condi ions wi h uncued ewa ds and no 100% con ingency be ween cues and ewa d. The app oach A senaul e al. 2013 ook is in e es ing, as hey aimed a a looking a he e ec s o ewa d in isual a eas independen ly o po en ially o he e ec s as a en ion. I a pa icula ask, ime o s imulus p edic s an upcoming ewa d hen a subjec will na u ally end o a end o he p edic i e cue. In a di e en BOLD MRI s udy, using a complex isual de ec ion ask, he e ec s o ‘incen i e’ in he isually e oked esponses, showed enhancemen o hose in se e al a eas including isual co ex (Engelmann e al. 2009). Incen i e was manipula ed by he amoun o u u e ewa d (money) o co ec ials; a guably no he same as ou ewa d manipula ion. Mo eo e , we also did no ha e a s imulus, which we use look o changes in isually e oked esponses. S ill i is concei able ha he changes we obse ed in ial esponses migh be o he same class as he ones men ioned in Engelmann e al. 2009. The e ec s o a en ion in BOLD MRI a e well known, wi h associa ed inc eases o BOLD esponse wi h a en ion (Jack e al. 2006, Syl es e e al. 2007, Donne e al. 2008, Pes illi e al. 2011). In humans a simila educ ion in BOLD MRI esponses in isual co ex o isual asks wi h he p esence o ewa d has been obse ed (Knapen e al. 2012). He e simila ly, ewa d was no p edic ed by any speci ic s imuli. In he p ima e, A senaul e al. 2013, obse ed as well ha a dec ease in MRI esponse wi h ewa d was ela ed o he ewa d amoun (highe ewa ds caused la ge dec eases in he MRI esponse). Using a pa adigm whe e di e en s imuli in he isual ield we e associa ed (o no ) wi h ewa d in human MRI BOLD expe i- men s, i was obse ed ha he esponses associa ed wi h ewa d s imuli had enhanced esponses ela i e o un ewa ded p esen a ions o he same s imuli (Knapen and Donne , pe sonal communica ion). Ou expe imen s we e pe o med in quasi o al da kness and e en hough we did no image in he co ical ep esen a ion o he o ea and ewa d had 100% p obabili y i 87 subjec s held ixa ion, we would no expec o ind he di e ence in esul s a ising om he ac we we e no eco ding he o ea ep esen a ion. These ambiguous esul s o he in luence o ewa d in co ical isual a eas as he subjec s pe o m isual asks, sugges ha ewa d migh be enhancing ewa ded s imuli ep esen a ion as well as educing he signal- o- -noise a io by means o educing he mean signal. Fu he esea ch is needed o be e unde s and his ques ion. 3.6 Task- ela ed Hemodynamic Response in Visual Co ex o an Audi o y-Mo o Task Finally, we wan ed o u he explo e he speci ici y o he ask- ela ed hemo- dynamic esponse o isual asks. On he ini ial assessmen o his p oblem in he labo a o y an audi o y/mo o ask was used, while eco ding in p ima y isual co ex. A ha poin esul s we e inconclusi e (i was no he ocus o ha s udy, bu see supplemen a y ma e ial o Si o in & Das 2009). We epea ed and expanded hose ini ial expe imen s. He e, we p esen da a (15 sessions o monkey T and 1 session o monkey S) o he same ask as ini ially used in Si o in & Das 2009. In his ask he animal ini ia ed ials by pulling a le e , a e a ixed delay (in he ange o 10-15 s, in di e en sessions) a one s a ed, las ing o a ixed du a ion (2-4 s, in di e en sessions) a e which, he one changes pi ch, cueing he animal ha i could elease he le e , he animal had a g ace pe iod (up o 2.5 s) o elease he le e , i i did elease i wi hin ha g ace pe iod, hen a ewa d was deli e ed and he ial ends. All his was done in da kness, he moni o is o . Animals ended o en ain o his ask and ini ia e a ial soon a e he end o he p e ious one. I esul ed ha hese ials had compa able du a ions o one ano he . The ixed ime in e al was selec ed so ha he du a ion o a comple ed ial o his mo o -audi o y ask was simila o he iming o he ixa ion ask p e iously used. 88 We p esen also da a om 7 sessions ( o monkey E) o a modi ica ion o his ask. Monkey E was ini ially ained on his same audi o y/mo o ask ha he o he 2 animals pe o med, bu wi h e y poo pe o mance. We modi ied he ask making i close o he ixa ion ask, and he animal’s pe o mance imp o ed. The da a p esen ed he e o monkey E is o his modi ied audi o y/mo o ask. In his modi ied ask, ials s a ed wi h he p esen a ion o a one, cueing he animal o pull a le e . The animal was gi en a g ace pe iod o pull he le e (up o 2.5 s), ailu e o do so, mean an abo ed ial. I he animal pulled he le e wi hin he g ace pe iod, hen he one changed pi ch and he animal was expec ed o hold o le e un il he one was u ned o (2-4 s, in di e en sessions), cueing he animal ha i could elease he le e . Once mo e he animal was gi en a g ace pe iod (up o 3 s) o elease he le e , and i i did elease i wi hin ha g ace pe iod, hen he animal was gi en ewa d and a ixed in e - ial in e al (10-12 s, in di e en sessions) s a ed un il he beginning o he ollowing ial. The da a p esen ed he e ( o ei he e sion o he audi o y/mo o ask) was om eco ding days whe e a leas one o he session was also eco ded. This o he expe imen was a ixa ion ask done in quasi- o al da kness. We aimed a eco ding ials in bo h sessions ( ixa ion ask o audi o y/mo o ask) wi h simila du a ions. The eco dings a e always om p ima y isual co ex ( o a subse o sessions, 12, elec ophysiological da a was also eco ded, da a no included he e). We compa ed he ial a e age hemodynamic esponse o each session o each ask ype: ixa ion o audi o y-mo o ask. The simila i y be ween he ask- ela ed esponses’ shapes o bo h ask ypes in V1 is s iking. In Figu e 3.9, a e wo eco ding examples Figu e 3.9a- is o monkey T on he o iginal audi o y/mo o ask (compa ed wi h he ixa ion ask) and Figu e 3.9g-l is da a o monkey E on he modi ied audi o y/mo o ask. Bo h examples we e compa ed agains he ixa ion ask. The hemodynamic ask- ela ed esponses we e simila be ween ei he o he audi o y/mo o asks, when compa ed o he ixa ion ask pe o med on he same day. 95 To explo e he ole ha ewa d is playing in he hemodynamic signal, we designed ano he a ia ion o he ixa ion ask whe e we manipula ed ewa d amoun . I on he one hand his ewa d- ela ed ask does no disambigua e he a o esaid ques ion, i unequi ocally sugges ed ha he ask- ela ed esponse is in luenced by ewa d. No only was he ampli ude o he ask- ela ed esponse modula ed by ewa d amoun , bu also a change in a slow, ial leng h, hemodynamic esponse e lec ed ewa d le el. Finally, in he las expe imen p esen ed, whe e a ixa ion ask was compa ed o an audi o y/mo o ask, lead us o conclude ha he hemodynamic ask- ela ed esponse can be a global esponse. A leas a ask ha did no engage isual co ex, as he one used, s ill elici ed a hemodynamic ask- ela ed esponse in p ima y isual co ex. Equally impo an ly, i was he s iking inding ha independen o he di e ences in he iming o he wo asks, in V1, he hemodynamic ask- ela ed esponse had on a e age a simila p o ile. Once mo e, his opens u he ques ions. I poin s o some ole o he decision o engaging on a ial, o some e o in doing o a ending o ials, independen o he speci ic de ails o he ask. The e is some hing in insically ela ed o he decision o aking up a ial ha migh be in luencing he exis ence o he hemodynamic ask- ela ed esponse. On a di e en analysis o he same ixa ion ask p e iously published in he labo a o y, we obse ed ha in ials ha he animal ails o ini ia e did no on a e age p esen a ask- ela ed esponse (Si o in e al. 2012, Figu e 6). These non-ini ia ed ials we e signi ican ly di e en om ini ia ed ones, independen ly o he subjec ecei ing o no a ewa d. In non-ini ia ed ials he subjec did no make an ac i e e o o engaging wi h ask on ha ial. In ials he animal did no comple e, bu ini ia ed, he di e ence o comple- ed ones, ewa ded, and he ones he animal ini ia ed bu does no comple e was no as compelling as he di e ence be ween comple ed ials and non- ini ia ed ones. 96 We looked back a he analysis and da a included in Si o in e al. 2012, Figu e 6. We calcula ed a Pea son’s co ela ion coe icien be ween each ial in a session and he a e age comple ed ial o ha session. Fo each session we ha e an a e age co ela ion o each ial ca ego y (non- ini ia ed, ini ia ed and ewa ded, o comple ed, and ini ia ed and no ewa ded, o abo ed). Figu e 3.11 shows a sca e plo o he a e age co ela ion be ween comple ed ials and non-ini ia ed o abo ed ones, o each session p e iously used o Figu e 6 in Si o in e al. 2012. This sugges ed ha independen ly o comple ing a ial he e was a signi ican di e ence be ween engaging o no in a ial. Figu e 3.11 – Lowe co ela ion be ween non-ini ia ed ials and comple ed ones, han be ween abo ed ials and comple ed ones. Sca e plo o he session by session co ela ion be ween comple ed ials wi h he a e age comple ed ial ( o each session) and he co ela ion be ween ei he abo ed (black) o non-ini ia ed ials (g ey) (n = 68 sessions, 2 animals, monkey S and T). E o ba s e lec s.e.m. Reg ession lines: g ey (sca e plo abo ed s. comple ed ials), y = 0.19x10-3 + 0.57x, 2 = 0.21, black (sca e plo non- ini ia ed s. comple ed ials), y = -0.28x10-1 + 0.14x, 2 = 0.04. The ask- ela ed hemodynamic ac i i y in V1 p obably e lec s di e en aspec s o beha io o he ha jus isual p ope ies o he scene. E en hough some o he expe imen s p esen ed he e we e es ing he in luence o ixa ion on he hemodynamic ask- ela ed esponse, i can be a gued ha he e ec s obse ed he e we e small and as obse ed in Figu e 3.5A a e age co ela ion (non-ini ia ed ials) (abo ed ials) 0 0.5 1 -0.5 0 0.5 1 a e age co ela ion (comple ed ials) 97 (dis ibu ion along he e ical axis) o each ixa ion du a ion p esen ed he e was a wide ange o hemodynamic esponse delays. This sugges ed ha ixa ion du a ion should no be igge ing he hemodynamic esponse (a leas as could be desc ibed by a simple hemodynamic esponse unc ion). The hemodynamic esponse, p obably e lec s many aspec s o beha io and physiology, we canno exclude he possibili y ha he e is a ole o ixa ion in his esponse, bu i should no be he sole d i e speci ically o he ask- ela ed hemodynamic esponse. The combined anecdo al e idence o a ew expe imen al a ia ions allows specula ion on wha mechanisms migh be (o migh no be) unde lying his hemodynamic ask- ela ed esponse. Bea ing in mind ha we expec his esponse o be global (p esen in isual co ex in esponse o an audi o y/ mo o ask, independen o he di e en iming o sequence o e en s o he audi o y/mo o ask ela i e o he ixa ion ask) and ha he e is a mo e p onounced di e ence be ween non-ini ia ed ials o comple ed ones han be ween abo ed ials and comple ed ones. I seems like a mechanism as a en ion could be e y ele an when conside ing his signal. Po en ially no a en ion in he sense o spa ial a en ion, when one would expec a leas some co ical dedica ion o he asks pe o med, bu a mo e global a en ion o an icipa ion o upcoming e en s. The e is e idence om esea ch in humans o global a en ion-like (e.g. Ress e al. 2000, Sil e e al. 2007, Syl es e e al. 2007, Donne e al. 2008, Pes illi e al. 2011) and ask- ela ed changes in BOLD MRI (Jack e al. 2006, Elkhe ali e al. 2015, G i is e al. 2015). We would be eage o see MRI expe imen s ha would look a o he b ain a eas (co ical and sub- co ical) ha could shed some ligh in o po en ial neu omodula o y mechanisms. One can also en e ain he idea ha he e is some po en ial o a global esponse ela ed o a ousal/engagemen o ha e a ole in his signal. Recen publica ions on he physiology o he locus ce uleus in he con ex o beha io , explo e he ole o phasic locus ce uleus i ing wi h he decision o execu e ac ions ha a e ask ele an (Kalwani e al. 2014 and 98 Joshi e al. 2016). Fo he simple asks ha we used, i migh be o eseeable ha he mos ele an decision ha equi es execu ion pe ains engaging in he ask. Wi hou u he expe imen s his emains in he ealm o specula- ion. Hence, ou empi ical obse a ions sugges ha he e is he po en ial ha he hemodynamic esponse is e lec ing he in luence o di e en neu omo- dula o y sys ems. Finally, we ha e ailed o ind neu al measu es ha eliably pa allel he hemodynamic esponse; he closes ela ed me ic, in hese obse a ions was hea a e, bu e en hea a e did no ully ack he obse ed nuances o he hemodynamic esponse. This e lec s p io wo k in he labo a o y whe e mul i-linea eg ession models combining a se o neu al eg esso s (MUA, a ious LFP bands) we e used o p edic he hemodynamic esponse (Si o in & Das 2010a). Such models we e only able o accoun o less han hal he a iance in he hemodynamic esponse. 99 4 Slow d i s in b ain blood olume a e associa ed wi h beha io al changes in p ima e V1 4.1 Abs ac Neu oimaging signals (e.g. BOLD MRI signals) a e complex and e lec bo h exogenous and endogenous esponses. The e is ex ensi e knowledge pe aining hemodynamic exogenous esponses, while endogenous espon- ses emain la gely elusi e. In he pas , mos s udies o neu oimaging ha e ypically ocused on sho - e m s imulus- ela ed hemodynamic changes on imescales o a ew seconds. He e, we ocus on he less in es iga ed long- e m luc ua ions in local b ain blood olume, and i s co ela ion o beha io - al engagemen . We assessed beha io al changes h ough spon aneous luc ua ions in he pe o mance o monkeys engaged in a pe iodic ixa ion ask. We simul aneously acqui ed hemodynamic signals ( h ough in insic signal op ical imaging), mul iuni neu onal ac i i y and local ield po en ials - LFP - ( h ough in aco ical mic oelec ode eco dings), hea a e (de i ed om he op ical imaging signal) and beha io al me ics ela ed o quiescen- ce. No ably, we obse ed obus inc eases in local b ain blood olume du- ing epochs in which he monkey was less engaged in he ixa ion ask, as compa ed o lowe blood olumes when he monkey was ac i ely engaged o mo e mo i a ed o pe o m he ask. Acco dingly, epochs o lowe engagemen we e associa ed wi h inc eased powe in lowe LFP equency bands, lowe i ing a es and lowe hea a e. The e o e, epochs o low b ain blood pe usion we e associa ed wi h highe hea a e. Finally, we compa ed he s a e-dependen slow d i in b ain blood olume wi h he hemodynamic ask- ela ed signal. The in e play be ween changes in blood olume and hea a e is complex; i on he one hand he slow d i in blood olume is nega i ely co ela ed wi h hea a e, he ial- ela ed changes a e posi i ely co ela ed. 100 4.2 In oduc ion The e is ex ensi e use o neu oimaging signals o in e ing neu al ac i i y. Techniques like blood-oxygen-le el dependen (BOLD) unc ional magne ic esonance imaging ( MRI) ha a e minimally in asi e and can measu e signals om he ull b ain ep esen an ad an age when compa ed o elec ophysiology. When alking abou neu oimaging signals one ends o ocus on signals d i en by ex e nal e en s. Howe e he e is signi ican e idence showing he p esence o endogenous esponses (like a en ional o ask- ela ed) in neu oimaging signals (Co be a e al. 1990, Too ell e al. 1998, Kas ne e al. 1999, Ress e al. 2000, Se ences e al. 2004, Jack e al. 2006, Syl es e e al. 2007, Donne e al. 2008) ha a e no co ela ed wi h local elec ophysio- logy (Si o in & Das 2009). The neu oimaging esponse is hen he esul o endogenous and exogenous con ibu ions. Exogenous con ibu ions e lec esponses o ex e nal s imuli (like isual s imuli in isual co ex) and ha e a linea ela ionship wi h local neu al ac i i y (Ca doso e al. 2012, Lima e al. 2014) and endogenous neu oimaging signals do no i ially e lec unde - lying changes in local neu al ac i i y (Si o in & Das 2009). Finally, in he isual sys em, he e is ample e idence o he p esence o bo h s imulus- ela ed and endogenous con ibu ions o he hemodynamic esponse, in human and non-human p ima es (Jack e al. 2006, Donne e al. 2008, Pes illi e al. 2011, Si o in & Das 2009). The e a e di e en ac i i y pa e ns ha e lec endogenous con ibu ions o he BOLD signal; in pa icula wo k in humans shows he p esence o BOLD modula ion wi h spa ial a en ion and ask s uc u e (Jack e al. 2006). In Jack and colleague’s s udy, subjec s we e asked o pe o m a nea h es- hold isual de ec ion ask and epo hei decision immedia ely o a e a ixed delay. They obse ed ha he BOLD signal had wo clea ly sepa able componen s: an inc ease in BOLD associa ed wi h spa ial a en ion o he nea h eshold s imulus, and ano he BOLD signal inc ease associa ed wi h 101 he s uc u e o he ask, and independen o he p esen ed s imulus. The second, he ask- ela ed signal, has also been obse ed in ou lab in non- human p ima es (Si o in & Das 2009). An endogenous signal ha has been widely epo ed in BOLD MRI s udies is associa ed wi h a en ion, he e a e se e al examples o a en ion in isual co ex (Kas ne e al. 1999 Some s e al. 1999, Gandhi e al. 1999, Ress e al. 2000, Ress & Heege 2003, Shulman e al. 2003, Se ences e al. 2004). P e ious wo k om ou labo a o y (Si o in & Das 2009) in non-human p ima- es pe o ming a pe iodic isual ixa ion ask compa ed ask- ela ed and s imulus- ela ed hemodynamic signals wi h elec ode eco dings, in p ima y isual co ex. Si o in and Das showed a ask- ela ed signal en ained o ial iming, in he absence o isual s imuli. This signal was compa able in mag- ni ude o some s imulus ela ed esponses and could no be p edic ed om local mul i-uni ac i i y (MUA) o local ield po en ials (LFP), unlike he s imulus componen . The e a e BOLD MRI s udies (Elkhe ali e al. 2015; G i is e al. 2015) as well ha show e idence o a ask-like ac i i y in human subjec s ha is dependen o pe o mance. In his s udy, subjec s we e asked o de ec a change in a s imulus’s empo al equency, in he con ex o isual, audi o y s imula ion, o bimodal audi o y- isual s imula ion (sub- jec s we e eques ed o epo changes in one o he senso y modali ies). In his ask, he e was e idence o a ‘ ask-ini ia ion ac i i y’ e lec ed in an inc ease in BOLD esponse in isual co ex wi h s a ing a block o ials when he subjec s we e asked o de ec a isual s imulus, ela i e o pe o ming he audi o y de ec ion ask. The au ho s obse ed as well sus ained changes in BOLD esponses. They epo ed an inc ease in baseline ac i i y wi hin a block o isual de ec ion ials ela i e o a block o audi o y de ec ion ials – ‘ ask-main enance ac i i y’. This las signal had highe magni ude in isual co ex o subjec s pe o ming be e on he isual de ec ion ask; hence pe o mance in luenced he BOLD MRI esponse. In he p esen s udy, we aimed a explo ing he po en ial co ela ion be ween di e en s a es o engagemen and he hemodynamic esponse. We explo- 102 ed he beha io al co ela es o he hemodynamic esponse in he ale animal p epa a ion, using he labo a o y’s es ablished echnique o simul a- neous unc ional neu oimaging (in insic-signal op ical imaging) and elec o- de eco dings om p ima y isual co ex o non-human p ima es pe o ming pe iodic isual ixa ion asks. The e is ex ensi e li e a u e ela ing changes in men al s a e o physiological changes. The e a e epo ed changes in hea a e and skin conduc ance concomi an wi h indices o a ousal. Malms om and colleagues ha e shown changes in hese me ics while subjec s we e wa ching a benign ilm (Malms om e al. 1965). E en ea lie wo k by Angelo Mosso in co ical pulsa ion showed changes in pulsa ion a es and p o iles wi h di e en men al s a es, o example sleep ( ansla ion o o iginal publica ion: Raichle & Shephe d 2014). The e a e also known changes in LFP wa ebands wi h a ousal/a en ion, in pa icula he ea ly wo k o Hans Be ge and he de elopmen o elec o- encephalog aphy. Hans Be ge showed ha alpha-band ac i i y inc eased when subjec s laid wi h hei eyes closed ela i e o when hey had hei eyes open ( o e iew see Mille 2001). Gi en he ex ensi e e idence o di e en ma ke s o endogenous ac i i y in he b ain we aimed a measu ing neu al and physiological signals wi h he goal o shedding ligh in o he unc ional ole o endogenous hemodynamic esponses. Speci ically, we eco ded: MUA, pai -wise MUA co ela ion, LFP and hea a e. As men ioned, he e a e physiological and neu al me ics ha cha ac e ize di e en aspec s o beha io al s a e. In his s udy we p opose ha he e is a signa u e o ask engagemen ha is e lec ed on he hemodynamic signal. To suppo ou claims, we also epo o he beha io al ma ke s. In he ask we ha e used, we looked a pe iods in which he subjec s we e engaged in he ixa ion ask, bu we also looked a pe iods in which hey we e disenga- ged, no comple ing ials. This wo k pa allels wo k done in a mo e es abli- shed beha io change: a ousal. We compa ed ou obse a ions wi h hose known o occu wi h changes in a ousal and ale ness. 103 A common physiological ma ke o a ousal is pupil diame e ( o a e iew see As on-Jones & Cohen 2005). Recen wo k (Reime e al. 2014, Vinck e al. 2015) looked a di e en s a es o a ousal in oden s (p obing locomo ion e sus quiescence) obse ing dis inc i e pa e ns in isual co ex: inc eased gain o isual esponses wi h inc eased a ousal, dec eased spiking noise- co ela ions, and e en ec ui men o di e en cell classes. In e ms o neu onal ac i i y, hey p esen ed e idence o changes in spiking pa e ns wi h a en ion and a ousal. A ousal was associa ed wi h imp o emen o signal o noise a io in co ical a eas wi h changes in ‘noise-co ela ions’ ac oss elec odes, e en hough inc eased a ousal was associa ed wi h lowe i ing a es (Li ings one & Hubel 1981). Changes in simila me ics a e associa ed wi h a en ion; compa ing co ela ions in he i ing pa e ns o indi idual cells on a ial by ial basis (‘noise-co ela ions’) i has been shown ha a en ion is associa ed wi h inc eases in i ing a e (Mo an & Desimone 1985, Luck e al. 1997), while a en ional imp o emen was associa ed wi h dec eases in in e neu onal co ela ions (Cohen & Maunsell 2009, Cohen & Kohn 2011). We ha e used in e neu onal co ela ions sligh ly di e en ly in his s udy, compu ing hose in ime as opposed o ac oss ial condi ions, which was mo e app op ia e o his s udy ( isual co ex is sensi i e o a ia ions d i en by eye mo emen s, and o pe iods he subjec s do no ixa e i becomes meaningless such compa isons ac oss engaged and disengaged ials, as he pa e ns o eye mo emen s can be dis inc ). Finally, he e is ecen e idence o changes in hemodynamic signal ha ma ch changes in a ousal, as measu ed by pupil dila ion (Pisau o e al. 2016). In his s udy hey looked a he hemodynamic esponse in he isual co ex o mice while iewing isual s imuli. They epo ed ha hemodynamic esponse had wo componen s; one ela ed o he isual s imuli he o he , mo e global and co ela ed wi h pupil diame e . The s udy we p esen he e pa allels Pisau o’s s udy and aims a u he explo ing he ask- ela ed componen o he hemodynamic esponse o global changes in hemodyna- 104 mic signal, in di e en s a es o engagemen in he ask. In ou s udy we did no use isual s imula ion, so we could educe in luence o isual s imuli in he eco ding egion. Mo eo e we include he e elec ophysiological da a no included in Pisau o’s s udy. We ained subjec s o pe o m a pe iodic ixa ion ask in quasi o al da k- ness, as epo ed be o e (Si o in & Das 2009). Animals pe o ming his ask na u ally exhibi pe iods o mo e and less engagemen in he ask. Ou pe o mance me ic uses he seg ega ion o ini ia ed e sus no ini ia ed ials o es ima e engagemen in he ask. We compa ed how he di e en measu ed signals ela e o pe iods in which he subjec was engaged in he ixa ion ask e sus pe iods in which i was no . Ou da a sugges s ha endogenous changes on he hemodynamic signal ha e a signi ican co ela- ion wi h beha io ; we obse ed inc eases in blood olume change wi h disengagemen in he ask. The changes in engagemen we e accompanied by changes in hea - a e; inc eased hea a e was associa ed wi h pe iods o highe engagemen . We ha e obse ed a endency owa ds an inc ease in lowe equency LFP powe and dec ease in highe equencies in mo e disengaged pe iods. The e was a sligh dec ease in i ing a es a mo e obus inc ease in i ing ‘noise co ela ions, ac oss elec odes, in mo e disengaged pe iods ela i e o mo e engaged ones. This s udy b ings a en ion in o he impo ance o accoun ing o endogenous signals’ con ibu ion o neu oimaging when using such echniques. 4.3 Me hods 4.3.1 Summa y In his chap e we used some da a p e iously used in o he publica ions (Si o in & Das 2009; Si o in e al. 2012) eanalyzed i o he ques ions add essed in his publica ion and collec ed a signi ican numbe o addi ional expe imen s. 111 alues o each session is exempli ied in Figu e 4.1b. Fo compa isons ac oss sessions, da a was z-sco ed, and he a e age aining and alida ion numbe s we e a e aged, wi h each session con ibu ing equally (Figu e 4.1d). To u he es he ela ionship be ween he alues o ini ia ed and non-ini ia ed ials, o each pai (co esponding o he mean esponse on each 20% o he da a o he wo classes) an angle was also calcula ed (ha ing ini ia ed ials as he e e ence). By plo ing he dis ibu ion o hese angles one can obse e how consis en o no is he ela ionship o ini ia ed and non-ini ia ed ials (Figu e 4.1c). 4.4 Resul s 4.4.1 Hemodynamic mean ial esponse co ela es in e sely wi h engagemen We wan ed o unde s and i slow changes in he hemodynamic esponse we e in luenced by beha io al signals, simila ly o p e iously epo ed (Pisau o e al. 2016) , in ou se up, in p ima y isual co ex (V1),. We se up o eco d esponses om non-human p ima es pe o ming a ixa ion, in quasi o al da kness, whe e only a ixa ion poin was p esen . We simul ane- ously eco ded in insic signal op ical imaging, ex acellula MUA and LFPs. The ixa ion ask he subjec s pe o med was a he simple, and wi h a limi- ed numbe o possible ou comes. Animals could hold ixa ion o a equi ed amoun o ime and ecei e ewa d; we call hese ials comple ed o e- wa ded. Animals could ini ia e he ixa ion ask, bu in e up ixa ion be o e he end o he equi ed ixa ion ime, no ecei ing ewa d: abo ed ials. O animals could no e en ini ia e he ixa ion ask: non-ini ia ed ials. The beha io s exhibi ed we e a he simple, bu adequa e o explo ing co ela es o ask engagemen in he hemodynamic signal. The e o e we ocused ou analysis on ini ia ed s. non-ini ia ed ials. We de ined a pe o mance me ic as he smoo hed p obabili y o ini ia ing a ial (see me hods). 112 The undamen al ela ionship be ween he animal’s pe o mance in a ask, hemodynamics, and di e en neu al and physiological measu emen s is illus a ed o a ep esen a i e session in Figu e 4.1. Blood olume changes inc ease when he subjec s we e less engaged in he ixa ion ask. This can be obse ed in Figu e 4.1 du ing pe iods o low engagemen (shaded g ay pe iods; pe o mance below 50%) he ial mean hemodynamic signal ends o be highe . To quan i y he ela ionship be - ween hemodynamic signal and pe o mance, we calcula ed he Spea man’s ank co ela ion be ween he mean ial hemodynamics and pe o mance ( = -0.531, p- alue < 0.01, in his session). In his session he e was a nega i e ela ionship be ween he animal’s engagemen and changes in blood olume; lowe engagemen was associa ed o highe olume. To unde s and i he obse ed changes we e eliable o his session, o ep esen ed an a e age endency bu we e no obse ed in sho e ime pe iods, wi hin each session; we used a 5- old c oss alida ion me hod. We spli each session in 5 con iguous segmen s; one segmen was kep as alida ion, he o he 4 comp ised he aining se . We compa ed he mean hemodynamic signal o ini ia ed e sus non-ini ia ed ials wi hin he es and alida ion se s (Figu e 4.1b). Fo his session we could seg ega e ini ia ed om non-ini ia ed ials. In pa icula when looking wi hin each o he 5 indi i- dual segmen s (see he g ay lines, which connec he a e age esponse o ini ia ed and non-ini ia ed ials in each segmen ) one could obse e ha in all ins ances he non-ini ia ed ials (in ed) had a highe hemodynamic signal han ini ia ed ones (in blue). We quan i ied his di ec ionali y o ini ia ed ials wi hin each segmen o 20% o he da a ha ing on a e age a smalle hemodynamic signal change han non-ini ia ed ials. This quan i- ica ion measu ed he angle be ween ini ia ed o non-ini ia ed ials. The measu e o his angle is ca ooned o one o he k- old segmen s on Figu e 4.1b. Angles wi hin he i s quad an indica e highe mean blood olume changes o non-ini ia ed han ini ia ed ials ( he e e ence we e he ini ia ed ials, and he same o all he o he measu es). 113 Pe o mance HemodynamicsMUA Spike-Spike Co LFP Alpha Gamma Hea Ra e a 0 100 Pe cen Ini ia ed -0.04 0.06 -dR/R 0 80 Fi ing a e (Hz) -0.2 0.5 Pea son’s -0.4 0.3 Hz -80 80 µV 0 5x10-5 powe 2x10-5 5x10-5 powe Sequence o ials 100 200 300 400 500 600 ARI 0.4 1.6 powe a io B ain Mo 0 0.8 pixel Eye P ob 0 1 P ob. eyes closed 114 Figu e 4.1 – How hemodynamic, hea a e and di e en neu al and physiological measu es ack changes o pe o mance in ixa ion ask. (a) co ela ion  = -0.352 -0.8 0 0.8 0 10  = 0.380 -0.8 0 0.8 0 10  = -0.200 -0.8 0 0.8 0 10  = -0.250 -0.8 0 0.8 0 10  = -0.101 -0.8 0 0.8 0 10  = -0.018 -0.8 0 0.8 0 10  = -0.158 -0.8 0 0.8 0 10  = -0.101 -0.8 0 0.8 0 10  = 0.408 -0.8 0 0.8 0 10 -0.8 0 0.8 0 10  = -0.572 26 44 26 44 0.02 0.18 0.02 0.18 -0.15 0 0.15 -0.15 0 0.15 -0.015 0 0.015 -0.015 0 0.015 b -8 8 -8 8 1.6 2.8 1.6 2.8 x10-5 x10-5 2.9 3.6 2.9 3.6 x10-5 x10-5 c e 90 270 180 0 49 -3 0 3 -3 0 3 aining se : 80% o da a alida ion se 20% o da a -0.8 0 0.8 0 10 co ela ion # sessions  = -0.297 Hemodynamics 30 -2 0 2 -2 0 2 -0.8 0 0.8 0 10  = 0.154 MUA 16 -1 0 1 -1 0 1 -0.8 0 0.8 0 10  = -0.202 Spike-Spike Co -1 0 1 -1 0 1 21 LFP -0.8 0 0.8 0 10  = -0.002 26 -1.5 0 1.5 -1.5 0 1.5 -0.8 0 0.8 0 10  = -0.006 Alpha 26 0-2 2 -2 0 2 -0.8 0 0.8 0 10  = -0.104 Gamma -3 0 3 -3 0 3 -0.8 0.80 0 10  = 0.331 Hea Ra e 68 0.85 1.1 0.85 1.1 0-1.5 1.5 -1.5 0 1.5  = -0.054 -0.8 0 0.8 0 10 ARI 25 -0.8 0 0.8 0 10  = 0.335 B ain Mo 92 -0.8 0 0.8 0 10  = -0.557 Eye P ob 0-2 2 -2 0 2 4 4 0-2 2 -2 0 2 4 6 4 6 34 0 1.8 0 1.8 0 0.05 0 0.05 alida ion se aining se d 115 Da a om a single expe imen al session ( o monkey T). Top: Animal’s pe o mance (see me hods) along a session as he p obabili y o ini ia ing a ial. Shaded g ey pe iods e lec pe iods in which he animal’s pe o mance was below 50%. Di e en aces: Mean hemodynamic esponse on each ial, o he same session as in a. No e he inc eases on he hemodynamic esponse du ing pe iods o lowe pe o mance (shaded g ay a eas). Simila analysis o he hemodynamic esponse o : hea a e, mul i-uni ac i i y elec ode (‘MUA’), spiking co ela ions be ween elec odes (‘spike-spike co ’), local ield po en ial (‘LFP’), powe o alpha-band LPF (8-12Hz), a ousal ela ed index (‘ARI’), “b ain mo emen ”, and eye s a us. (b) Wi hin session c oss- alida ion o he hemodynamic esponse, o he same session as in a. Ve ical axis: a e age esponse o 20% o he da a (in successi e segmen s) in ini ia ed ials (blue) e sus non-a emp ed ials ( ed); on he ho izon al axis is he a e age esponse o he emaining 80% o he da a (once mo e seg ega ed by ini ia ed and non-ini ia ed ials, espec i ely). The lines connec ini ia ed e sus non-a emp ed ials in each 20% segmen . No e, wi hin each colo ca ego y, he e is less dispe sion on he ho izon al han e ical axis (as o he wise expec ed, ha esponses o 20% o he da a a e mo e a iable be ween one ano he , han esponses o 80% o he da a). The angle measu emen will be used in la e analysis (e); he me ic used conside s he slope o he line ha connec s ini ia ed o non- a emp ed ials. (c) S acked his og am o he Spea man’s co ela ion be ween hemodynamic esponses and pe o mance o each session. Di e en shaded egions indica e he con ibu ion o each o he h ee monkey (ligh e g ey: S, in e media e g ey: T, da ke g ey: E). (d) Simila o c bu o he subse o sessions ha had p- alue associa ed wi h Spea - man’s co ela ions < 0.05, no e ha mos endencies ge mo e e idenced. (e) Simila o panel b, o he popula ion o da a (no single session), da a p esen ed as z-sco e. Di e en animals a e coded wi h di e en symbols (S: ci cle, T: diamond, E: squa e). The colo ed c osses indica e he mean o each condi ion ( o cla i y sake i is no included he means o each animal, bu hey a e on a e age consis en wi h he global mean). ( ) Popula ion measu e o he angula dis ibu ions o he slopes be ween he neu al esponses om ini ia ed o non-a emp ed (indi idual angles as in panel b). Di e en colo codes o he di e en animals as in panel c. The session p esen ed is ep esen a i e o he popula ion, as no ed when looking a he summa y o he popula ion da a, Figu e 4.1c- . The dis ibu ion o Spea man’s ank co ela ion o he di e en sessions was on a e age 116 nega i e o all 3 animals. I indica es a clea endency o inc eased blood olume du ing lowe pe o mance pe iods (Figu e 4.1c, e en mo e o he subse o sessions ha had eached a signi ican Spea man’s co ela ion, Figu e 4.1d); hese esul s a e ein o ced by looking a he popula ion o z- sco ed 5- old c oss alida ion o all session, Figu e 4.1e, and he angula di ec ionali y esul ing om he same analysis, Figu e 4.1 . 4.4.2 Complex in e play be ween hemodynamics and hea a e When looking a neu oimaging signals, one o he ele an me ic is hea a e. In heal hy subjec s, he a e a which new esh blood a i es will be in ima ely linked o ca diac ou pu . In he con ex o beha io he e a e se e al known hea a e modula o y e ec s: es ing is also associa ed wi h dec eased hea a e, ca diac decele a ion in an icipa ion o beha io al e en s; s a le esponses associa ed wi h inc eased hea a es, b adyca dia associa ed wi h a en ion, o name a ew ( o a summa y o ea ly esul s see e.g., Lacey & Lacey 1970, Lacey & Lacey 1978). We he e o e looked a hea a e in pe iods o highe and lowe engage- men , simila ly o he analysis p esen ed o he hemodynamic esponse. On Figu e 4.1a i can be no iced ha pe iods o lowe engagemen in he ixa ion ask we e associa ed wi h dec eases in hea a e ( = 0.384, p- alue < 0.01, o his single session, and i had he same end o he popula ion, as can be no iced in Figu e 4.1c- ). This sugges ed a coun e in ui i e an i-co ela ed ela ionship be ween he mean ial hemodynamic signal and mean ial hea a e. One o he aspec o he ela ionship be ween hemodynamics and hea a e was ha he co ela ions ended o be mo e nega i e o sessions wi h lowe o e all engagemen o he animal, as no ed in Figu e 4.2a. The changes in mean ial hea a e wi h disengagemen ma ch ou ex- pec a ion. Simila ends can be obse ed in a ousal; inc eases in a ousal a e associa ed wi h inc eases in hea a e, con e sely dec eases in a ousal 117 come associa ed wi h dec eased hea a e. In wha conce ns slow changes in blood olume he obse ed esul s we e mo e unexpec ed. Ea lie obse a ions om he lab (Si o in & Das 2009) o hea a e changes du ing a ial showed concomi an changes wi h he hemodynamic esponse – no an i-co ela ions we e ound in ha s udy. To con i m ha he ela ionship be ween sho e m changes in hea a e and hemodynamics we e posi i ely co ela ed e en when looking a di e en s a es o engagemen , we looked a he a e age dynamic esponses o bo h signals in a sub second scale (we looked a he signals a he imaging equency). We looked a he a e age esponses o ini ia ed and non-ini ia ed ials. No e ha non-ini ia ed ials a e a cons uc om he way we p esen ials o he subjec . In ou pa adigm, ials we e p esen ed pe iodically, independen- ly o he animal engaging on hem o no . The e o e, we ecognize ha his compa ison can disp opo iona ely a e age ou con ibu ions wi hin non- ini ia ed ials, as he e is no e en ime s uc u e. We obse ed ha he a e age hemodynamic esponse o ini ia ed ials had a peak a ew seconds in o he ial (Figu e 4.2b). This esponse was o e all lowe o ini ia ed han o non-ini ia ed ials. Simila ly, he a e age hea a e esponses o ini ia ed ials had also a peak (ea lie han he peak in he hemodynamic esponse, Figu e 4.2c). I should be no ed ha he a e age hemodynamic esponse change o comple ed ials (a subse o ini ia ed ials; he ones ha we e ewa ded) co esponds o he p e iously desc ibed ask- ela ed hemodynamic esponse (Si o in & Das 2009). Mo eo e , hese obse a ions a e consis en wi h ou ea lie obse a ions (Si o in & Das 2009). I is wo h no ing ha he mean hea a e o non-ini ia ed ials was lowe han o ini ia ed ones (as al eady seen in Figu e 4.1). This sugges s a posi i e co ela ion be ween changes wi hin a ial in hea a e and hemodynamics, on op o an i-co ela ed changes in slowe ime scales, slowe han he scale o indi idual ials, be ween he same me ics. To es ablish i he co ela ion be ween he hemodynamic and hea a e signals was in he ull ime se ies domina ed by he as e ime scales 118 (posi i e co ela ion) o by he slowe ime scales (nega i e co ela ion) we compu ed he c oss-co ela ion o each session in wo condi ions. In he i s , we simply calcula ed he Pea son’s co ela ion coe icien be ween he ull ime se ies o changes in hemodynamic and hea a e. This esul ed in a p e e en ially nega i e co ela ion (Figu e 4.2d). On he o he hand, in he second app oach we discoun ed he slow changes in he wo me ics, by sub ac ing o he ime cou se o each ial he mean esponse o ha ial. In his case he co ela ion was p e e en ially posi i e (Figu e 4.2e). Figu e 4.2 – Rela ionship be ween he hemodynamic esponse and hea a e. (a) Session by session, Spea man’s co ela ion be ween he mean ial hemodynamic esponse and he mean ial hea a e esponse as a unc ion o a e age pe o mance on he session (open ci cles: monkey S, open dia- monds o hombus: monkey T, open squa es: monkey E). (b) T ial igge ed a e age o he magni ude no malized hemodynamic esponse. In blue, he a e age esponse o ini ia ed ials, in ed, he esponse o non-ini ia ed ials ( he di e en ones indica ed di e en animals, ligh e blue and ligh e 0 0.5 1 -1 -0.5 0 0.5 1 Spea man's co ela ion Pe o mance 0 5 10 15 -0.6 0 0.6 Hemo ime (s) no m -dR/R Hea Ra e 0 5 10 15 -0.6 0 0.6 ime (s) no m HR a b c -0.8 0 0.8 0 10 co ela ion # sessions = -0.157 20 -0.8 0 0.8 0 10 co ela ion # sessions = 0.271 20 C oss co ela ion (mean ial sub ac ion) C oss co ela ion (no mean ial sub ac ion) d e 119 ed indica e animal S, in e media e ones, monkey T, and da ke ones, monkey E). (c) Simila o b bu o hea a e changes. (d) Peak c oss co - ela ion coe icien o he ull hemodynamic ace and ull hea a e ace (only posi i e lags conside ed). (e) Simila o d, bu whe e o he hemo- dynamic and hea a e aces he mean ial ( o each ial) we e sub ac ed. The ela ionship be ween hemodynamic signals and hea a e changes is o complex na u e, and no linea . On a as ime scale (sub- ial ime du a ion), hey a e co ela ed; he e en s happening wi h in a ial migh igge changes in hea a e ha will ha e a e lec ion in he same di ec ion in he hemodynamic esponse. On a slowe ime scale, he wo me ics ended o be an i-co ela ed. P edic ing he hemodynamic esponse based on hea a e is he e o e no i ial: ma hema ically he wo me ics ha e a non- linea i y ela ionship. 4.4.3 Di e en neu al me ics and hei co ela ion o engagemen To unde s and he neu al co ela es o he hemodynamic changes we simul aneously eco ded MUAs and LFPs wi h he imaging esponse (in he same c anial window). Fi ing a es ended o dec ease in he pe iods he animal was less engaged in he ixa ion ask: he e was a weake and posi i e co ela ion be ween i ing a es and pe o mance (Figu e 4.1a,b,  = 0.171, p- alue < 0.01, o MUA). We specula e ha in pe iods he animal was less engaged in he ask, he subjec could ake mo e b eaks and close i s eyes mo e o en and o longe pe iods educing e en u he he ain e ec s o ha ing isual s imula ion om ixa ion ( i ually, he only sou ce o ligh in he oom). On he o he hand, we obse ed he e e se pa e n o he co ela ed ac i i y ac oss elec odes (see me hods o ou de ini ion o he co ela ed ac i i y be ween elec odes). In mos o he sessions ha we had elec ophysiological eco dings, we had wo elec odes in isual co ex a ew millime e s apa . This spike-spike co ela ion was on a e age highe o pe iods when he animal was less engaged in he ask (Figu e 4.1a,b ‘spk-spk-co ’:  = -0.366, p- alue < 0.01). I should be no ed ha his ela- ionship is in e se o wha happens wi h spiking, and so he inc eases in 120 spike-spike co ela ion a e no he esul o o e all inc eases in i ing a es. The ela ionship be ween in e neu al co ela ions and pe o mance could ha e been expec ed om wha is known o a ousal and ‘noise co ela ions’ (Li ings one & Hubel 1981, Cohen & Maunsell 2009, Cohen & Kohn 2011). Local ield po en ials ( he ex acellula po en ial in he icini y o he elec o- des) had no consis en ela ionship wi h pe o mance on he ask (Figu e 4.1a,b, LPF:  = 0.013, p- alue = 0.7) and he same lack o ela ionship held ue o he popula ion (Figu e 4.1c- ). We looked a di e en wa e-bands o he LPF signal, alpha is epo ed he e, gi en i s ele ance in he a ousal li e a u e. Alpha powe (7-12 Hz) had a nega i e co ela ion wi h pe o mance (alpha:  = -0.083, p- alue = 0.04, o he session in Figu e 4.1a,b). Ou obse a ion ha alpha powe inc eased in pe iods ha animal pe o ms poo ly is consis en wi h p e ailing ideas o igilance/a en ion li e a u e (e. g. Da ies & K ko ic 1965). In ou expe imen , p esumably in less engaged pe iods he animal was also less a en i e. Simila o he obse a ions in alpha, gamma band powe (30-90Hz) had a simila p o ile (gamma:  = -0.340, p- alue < 0.01, o he session in Figu e 4.1a,b). This was unexpec ed, alpha and gamma end o be an i-co ela ed. Conce ning ial an icipa ion he e is e idence o gamma inc eases ela i e o condi ions o low expec a ion (Lima e al. 2011). We he e o e looked a one o he me ic, ‘a ousal ela ed index’ (named and desc ibed elsewhe e: Moo e e al. 2014, see me hods o de ails). Summa ily, his index acks he endency ha in sleep one obse es inc eases in low equency powe and dec eases in he high equency powe . Ou esul s indica e ha his index does no ully ack pe o mance (ARI:  = -0.166, p- alue < 0.01, o he session in Figu e 4.1a,b). I is wo h men ioning ha occipi al co e age as he one we had o V1 imaging expe imen s is no he app op ia e loca ion o iden i y sleep episodes. Ne e heless, we do no expec hese esul s o e lec mos ly sleep, as he popula ion da a in Figu e 4.1c- shows a signi ican 127 able o ack some o he changes in b ain s a es. These come accompanied by changes in in a-co ical co ela ions, hea a e, and quiescence. Finally i emains open wha he bes neu al co ela e o endogenous changes in hemodynamic esponse migh be. 4.6 Acknowledgmen s We would like o hank E ic DeWi o he ui ul discussions. 128 129 5 Gene al Discussion Neu oimaging is bo h ex emely use ul and also a a he complex me hodo- logy o in e p e . In his hesis he aim was o u he explo e neu oimaging in he con ex o beha io . We di ided he wo k p esen ed he e in o wo b oad pa s. The i s , aimed a add essing he ela ionship be ween s imuli- - ela ed hemodynamic esponses and unde lying neu al ac i i y, in ea ly senso y a eas (Chap e 2). The second aimed a explo ing he ep esen a- ion o beha io - ela ed endogenous esponses in he hemodynamic espon- se, and also neu ally, as well, in ea ly senso y a eas (Chap e s 3 and 4). The wo k p esen ed he e does no close any o he linge ing ques ions in he neu oimaging ield. I ne e heless succeeds in b inging mo e a en ion o he ques ion o how o analyze esponses happening a di e en ime sca- les, and coming om di e en sou ces. Is epea ing ials mul iple imes and a e aging hei esponses he bes way o cap u e ask ele an ea u es? 5.1 Exogenous esponses in V1 In Chap e 2 we discussed how s imulus- ela ed and ask- ela ed compo- nen s can be dis inguished in he neu oimaging signal when looking a ea ly senso y a eas. We obse ed ha in a pe iodic ask he e was a s imulus- - ela ed con ibu ion o he hemodynamic esponse and a ask- ela ed com- ponen . Mo eo e hese wo componen s added linea ly o gi e he esul ing hemodynamic esponse. This ela ionship hough simple did no ha e o be like his, we could ha e ound e idence o in e ac ion be ween he s imulus and ask componen s. Finally, he ela ionship be ween he hemodynamic esponse and unde lying neu onal ac i i y has also been subjec o signi i- can deba e ( o a e iew see: Logo he is & Wandell 2004). The p e ailing idea has been ha he hemodynamic esponse mos ly e lec s me abolic de- mand, and ha i is be e acked by LFP ac i i y. The e is e idence in his di ec ion (e. g., Logo he is e al. 2001, Niessing e al. 2005, Goense & 130 Logo he is 2008). The wo k p esen ed he e sugges s also ha spiking ac i i- y is well co ela ed wi h he s imulus ela ed hemodynamic esponse. Mo e ecen wo k om he lab showed ha o he same ask (whe e we e included se e al o he same da ase s p esen ed on Chap e 2) spiking is a be e p edic o o he s imulus ela ed hemodynamic esponse han LFP (Lima e al. 2014). Thus we ha e good e idence ha exogenous hemo- dynamic esponses in V1 a e linea ly ela ed o changes in local spiking ac i i y. The e o e decon ol ing he s imulus- ela ed hemodynamic espon- se wi h an HRF p o ides a good es ima e o unde lying neu al ac i i y, pa i- cula ly ele an o MRI s udies. This also sugges s ha he hemodynamic esponse is igh ly linked o spiking ac i i y and p esynap ic ac i i y in o a senso y a ea, as anyway obse ed in di e en expe imen s (Gu den e al. 2006, Lee e al. 2010, Sco & Mu phy 2012, Kahn e al. 2013). Finally, o he neu oimaging communi y, his ein o ces he impo ance o app op ia e ask design: aiming a es ima ing ask non- ele an con ibu ions (in ou case, we we e in e es ed in cha ac e izing s imulus ep esen a ion) in o de o sepa a e hose om s imulus- ela ed componen s ha will ha e a linea ela ionship o unde lying local neu al ac i i y, as leas in ea ly senso y a eas. The ques ion o wha he neu onal na u e o endogenous hemodyna- mic esponses is emains unanswe ed. 5.2 Endogenous esponses in V1 In Chap e 3 we aimed a cha ac e izing beha io aspec s o he p e iously desc ibed ask- ela ed hemodynamic signal (Si o in & Das 2009). In he o iginal publica ion his signal had been cha ac e ized as p esen in V1 in a pe iodic ixa ion ask, and wi h a ime cou se ma ching ha o he ial du a- ion. We in es iga ed his issue u he by manipula ing ial du a ion and ixa ion du a ion. The hemodynamic esponse ends o ack he du a ion o he ials, bu i s ime cou se poin s o a ial ini ia ion esponse. As we saw ha al e na ion be ween wo schedule egimes did no p oduce signi ican 131 changes o a la ge po ion o ime in o long ials, when compa ed o sho ones. Then we looked a he e ec o ixa ion du a ion. The ask- ela ed esponse keeps acking he o al du a ion o he ial, bu longe ixa ions end o be associa ed wi h highe ask- ela ed esponse ampli udes and also slowe in he peaking ime. Taken oge he , his sugges s ha he hemo- dynamic ask- ela ed esponse is in luenced by he du a ion o ials. Impo an ly i migh be igge ed by ini ia ion o a ial, bu modula ed by he ime he animal is equi ed o hold ixa ion. These a e a he speci ic ea u- es associa ed wi h he speci ic ixa ion ask we used. We used a di e en ask, no elian on ixa ion o add ess his ques ion. We used a ask ha used audi o y s imuli as cues and mo o esponse and ou pu . Fi s , we obse ed a obus ask- ela ed esponse in V1 o his ask, wi h compa able magni udes. We hen looked a he ime cou se o such esponses be ween he wo ask ypes and ound g ea simila i y be ween hose. Mo eo e , i should be once mo e poin ed ou ha we ha e used wo sligh ly di e en e sions he audi o y/mo o ask, wi h simila esul s on bo h. I is wo h no ing ha he wo a ia ions di e signi ican ly in hei e en sequence. One ask was designed o emula e, as much as possible, he ixa ion ask, bo h in i s e en sequence ( ials we e p esen ed au oma ically o he animal ha could engage on hem o no ), and iming. The o he ask a ia ion was di e en : he animal ini ia ed he ials, and he e was a hia us o se e al seconds be ween ial s a and he audi o y cue coming on, and inally ewa d. I is s ill ema kable he simila i y in he ime cou se be ween his audi o y/mo o ask and he ixa ion ask. This sugges s ha he e is a po en ial ole o ini ia ing a ial (engaging in he ask) as a igge ing signal o he esul ing hemodynamic ask- ela ed ac i i y. We also looked a he e ec s o ewa d/mo i a ion o comple ing ials, and obse ed ha ewa d amoun inc eased he p obabili y o ini ia ing a ial and i inc eased he magni ude o he ask- ela ed hemodynamic changes. We obse ed as well a slowe end on he hemodynamic signal. This end had he unexpec ed p ope y o dec easing as one goes in o a block o high 132 ewa d (and inc easing when p og essing in o a low ewa d block). Apa om hese changes, we ha e no ound signi ican changes in he shape o he empo al p o iles o high o low ewa d ials. P e ious wo k in he lab (Si o in e al. 2012) had al eady poin ed ou ha he s a kes di e ence lay be ween ini ia ed o non-ini ia ed ials. A some le el his is i ial: i a ask- ela ed signal; no ask, no signal. Bu he ques ion emains; wha in he ask igge s his signal? Is i e o ? Is i ewa d? Is i expec a ion? Is i some hing else? Unde s anding he beha io o such sig- nal migh help hin a wha mechanisms unde lie i . In he inal chap e , Chap e 4, we explo ed u he he idea o engagemen o a ousal and i s in luence on he hemodynamic esponse. We no iced ha he slow changes in he hemodynamic esponse had signi ican ampli ude and hey ended o inc ease as engagemen in he ask dec eases, simila ly o he obse a ions by Pisau o and colleagues (Pisau o e al. 2016). Ou wo k he e aimed a u he con i ming he non-local o igin o hese hemodynamic changes. We looked a se e al co ela ions be ween he di e en measu emen s, and in pa icula we no iced a non-linea ela ionship be ween hea - a e and hemodynamic esponses; as i he e we e wo dis inc coupling modes ope a ing in wo di e en ime scales. This obse a ion migh be pa icula ly ele an when conside ing how o discoun hea - a e om he MRI signals ( o some discussion abou hea a e con ibu ions o BOLD MRI and how o accoun o hose see Chang e al. 2009). In summa y, he e is a ask- ela ed signal; i is p esen in V1 o isual asks as well as o an audi o y/mo o asks, i is mo e global han a modali y speci ic signal; i is p esen in ini ia ed ials independen ly i ials we e e- wa ded o no ; and he e is a slowe ( ial leng h) signal ha acks p obably ask engagemen o a ousal. I would be ele an o explo e he ole o neu- omodula ion, wi h his e idence we sugges i would be ele an o look a no epineph ine in pa icula . 133 5.3 Fu u e di ec ions 5.3.1 Locus coe uleus – no ad ene gic sys em No epineph ine (NA) and locus coe uleus (LC) ac i a ion a e an ex emely in e es ing sys em o u he specula ion o a neu omodula o y con ibu ion o he hemodynamic signal ela ed o ask engagemen . The wo k by As on- Jones and colleagues showing he wo modes o ac i a ion o LC, onic and phasic, and i s ela ionship o beha io is a he ele an ( o a e iew see As on-Jones & Cohen 2005). Tonic discha ge associa ed wi h global changes in ask disengagemen and pe o mance, and phasic changes associa ed o ou come and ask ele an ea u es like ewa d ela ed a ge s (e.g. As on-Jones e al. 1994, Ushe e al. 1999). Mo eo e , mo e ecen wo k by Gold and colleagues (Kalwani e al. 2014, Joshi e al. 2016) whe e hey obse ed ha LC (phasic) ac i a ion was ela ed o ‘ac i e’ ac ions di e en ly han o wi hholded ones. E en hough hese wo condi ions di e in he ac ions he subjec had o pe o m bo h lead o a ewa ded ou come; sugges ing ha LC phasic ac i a ion encodes goal-di ec ed ac ions mo e han solely ewa d. I would be ex emely in e es ing o eco d ac i i y di ec ly om LC in bo h he ixa ion ask and he audi o y/mo o ask, and, simul aneously, obse e no only LC ac i i y changes du ing his ask, bu also how LC esponses when he animal goes in o low pe o ming pe iods as we obse ed in Chap e 4. Neu oimaging o such an a ea wi h OIIS is no possible gi en i s loca ion in he b ain. The use o MRI echniques can be challenging gi en he loca ion o LC, i s small size o he ac ha b ains em expe iences signi ican pulsa ion. The use o app op ia e coils could aid in his ask, bu wi h he disad an age ha i could be ha d o esol e he whole b ain. Finally, o my unde s anding, he exac pa e n o ac i a ion/deac i a ion associa ed wi h he known elec ophysiological esponses o LC is no well cha ac e ized in neu oimaging. Mo eo e , MRI echniques a e no ideally posed when looking a slow changes (in he scale o se e al seconds o 134 minu es) in esponse gi en limi a ion in main aining he magne ic ield in he scanne . Ideally, expe imen s in ol ing LC would use elec ophysiology. Al e na i ely, pha macology could be used while eco ding he same hemo- dynamic esponses wi h OIIS, in he same asks we used be o e. Does inac- i a ion o NA ecep o s in luence he hemodynamic esponse? Visual co ex ecei es LC inne a ions (e.g. Koso sky e al. 1984). As a no e, ini ial e- sea ch on NA modula ion in isual co ex was associa ed wi h isual plas ici- y ( o a b ie e iew, see Silli o 1986). Mo e ecen wo k wi h sys emic inhibi ion o NA anspo e s has been linked o changes in he ‘de aul mode ne wo k’ (DMN, on es ing s a e MRI and he DMN see Biswal e al. 1997, Raichle e al. 2001, espec i ely) (Minzenbe g e al. 2011). In oden s, NA was shown o ha e a ole in as oglial esponses (Pauke e al. 2014). In hese expe imen s hey ha e looked a as ocy ic Ca2+ ac i i y in isual co ex. This ac i i y had low baseline le els, bu pa ing o a isual s imulus wi h pe iods o locomo ion (whe e NA in isual co ex is ele a ed), consis- en ly inc eased as ocy ic Ca2+ ac i i y. In he con ex o hemodynamic esponses, hese esul s a e e y in e es ing. In a nu shell, NA modula ion in o isual co ex is a p omising a enue o pu sue ollowing he wo k p esen- ed in his hesis. 5.3.2 Neu al basis o neu oimaging The wo k p esen ed in his hesis sheds ligh on he in e p e a ion o neu oimaging signals. Mos ques ions pe aining o i , howe e , s ill emain. In he in oduc ion he e was an e o o b inging o ligh he unknown na u e o he de ini i e mechanisms unde lying he hemodynamic esponse. The use o in insic signal op ical imaging allowed he p esen a ion o esul s in his hesis. Mo ing o wa d he e a e se e al esea ch ques ions ha migh be be e add essed wi h he use o o he expe imen al echniques. To unde s and and dis inguish me abolic and ascula con ibu ions o he esul ing hemodynamic signal i would be ele an o use echniques whe e 135 hose would be sepa a ed. O pa icula in e es a e me abolic measu es ha can ha e a mo e di ec ela ionship wi h unde lying neu onal ac i i y. We p opose ha a echnique which sha es simila i ies wi h in insic signal op ical imaging would be ele an o ad ance ou unde s anding o neu o- imaging: au o luo escence imaging, speci ically, la op o ein luo escence imaging. I is based on changes wi h oxida i e espi a ion in mi ochond ia o molecula species ha luo esce di e en ly in hei oxidized and educed o m (Chance e al. 1979). These a e e en mo e in e es ing because hey luo esce in he isible spec um. No only ha e such echniques been used success ully in oden s (e.g. Shibuki e al. 2003) as hey ha e p e iously been used in he lab (Si o in & Das 2010b). This echnique simila ly o OIIS does no equi e he use o any agen like dyes o con as o eco d he esponse. This ep esen s an ad an age, as hose a e o en oxic and equi e injec ion/ applica ion. I would be in e es ing o epea all expe i- men s wi h bo h OIIS and au o luo escence imaging. O pa icula ele ance would be o compa e he ela ionship be ween he wo imaging me hods o exogenous and endogenous con ibu ions o he hemodynamic esponse. Is he hemodynamic-me abolic coupling p ese ed o exogenous and endoge- nous signals? I ha e de o ed li le a en ion o he elec ophysiological eco dings in he collec ed da a. We consis en ly collec ed MUA da a, bu i s ela ionship wi h he hemodynamic esponse was no ex emely in o ma i e, o he han, ein- o cing ha he endogenous esponses do no ha e hei o igin in V1. This in o ma ion is ele an and aluable pe se, in my opinion i absolu ely s a es he impo ance o looking in o o he b ain egions. Howe e i con ibu es o he unde s anding o neu o ascula coupling in a limi ed way. I would be ele an o do mo e g anula analysis o he elec ophysiological da a, which could po en ially b ing o ligh mo e in o ma ion. 136 Re e ences Ai ken, P.G., Fayuk, D., Somjen, G.G. & Tu ne , D. a, 1999. Use o in insic op ical signals o moni o physiological changes in b ain issue slices. Me hods (San Diego, Cali .), 18(2), pp.91–103. Alb ech , D.G. & Hamil on, D.B., 1982. S ia e co ex o monkey and ca : con as esponse unc ion. Jou nal o neu ophysiology, 48(1), pp.217– 37. A ieli, A., G in ald, A. & Slo in, H., 2002. Du al subs i u e o long- e m imaging o co ical ac i i y in beha ing monkeys and i s clinical implica ions. Jou nal o Neu oscience Me hods, 114(2), pp.119–133. A ieli, A., Shoham, D., Hildesheim, R. & G in ald, A., 1995. Cohe en spa io empo al pa e ns o ongoing ac i i y e ealed by eal- ime op ical imaging coupled wi h single-uni eco ding in he ca isual co ex. Jou nal o neu ophysiology, 73(5), pp.2072–2093. A senaul , J.T., Nelissen, K., Ja aya, B. & Vandu el, W., 2013. Dopamine gic Rewa d Signals Selec i ely Dec ease MRI Ac i i y in P ima e Visual Co ex. Neu on, 77(6), pp.1174–1186. As on-Jones, G. & Cohen, J.D., 2005. An In eg a i e Theo y o Locus Coe uleus-No epineph ine Func ion: Adap i e Gain and Op imal Pe o mance. Annual Re iew o Neu oscience, 28(1), pp.403–450. As on-Jones, G., Rajkowski, J., Kubiak, P. & Alexinsky, T., 1994. Locus coe uleus neu ons in monkey a e selec i ely ac i a ed by a ended cues in a igilance ask. The Jou nal o neu oscience : he o icial jou nal o he Socie y o Neu oscience, 14(July), pp.4467–4480. A well, D., Buchan, A.M., Cha pak, S., Lau i zen, M., Mac ica , B. a & Newman, E. a, 2010. Glial and neu onal con ol o b ain blood low. Na u e, 468(7321), pp.232–43. A well, D. & Laughlin, S.B., 2001. An ene gy budge o signaling in he g ey ma e o he b ain. Jou nal o Ce eb al Blood Flow and Me abolism, 21(10), pp.1133–1145. Bashinski, H.S. & Bacha ach, V.R., 1980. Enhancemen o pe cep ual sensi i i y as he esul o selec i ely a ending o spa ial loca ions. Pe cep ion & psychophysics, 28(3), pp.241–248. Bi n, R.M., Smi h, M.A., Jones, T.B. & Bande ini, P.A., 2008. The espi a ion esponse unc ion: The empo al dynamics o MRI signal luc ua ions ela ed o changes in espi a ion. Neu oImage, 40(2), pp.644–654. Biswal, B.B., Van Kylen, J. & Hyde, J.S., 1997. Simul aneous assessmen o low and BOLD signals in es ing-s a e unc ional connec i i y maps. NMR in biomedicine, 10(4-5), pp.165–170. Boly, M., Bal eau, E., Schnake s, C., Degueld e, C., Moonen, G., Luxen, A., Phillips, C., Peigneux, P., Maque , P. & Lau eys, S., 2007. Baseline b ain ac i i y luc ua ions p edic soma osenso y pe cep ion in humans.