The Regulatory Role of the Human Mediodorsal Thalamus

HAL Id: hal-02341974

https://hal.archives-ouvertes.fr/hal-02341974

Submitted on 10 Feb 2021

HAL is a multi-disciplinary open access

archive for the deposit and dissemination of

sci-entific research documents, whether they are

pub-lished or not. The documents may come from

teaching and research institutions in France or

abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est

destinée au dépôt et à la diffusion de documents

scientifiques de niveau recherche, publiés ou non,

émanant des établissements d’enseignement et de

recherche français ou étrangers, des laboratoires

publics ou privés.

The Regulatory Role of the Human Mediodorsal

Thalamus

Giulio Pergola, Lola Danet, Anne-Lise Pitel, Giovanni A. Carlesimo,

Shailendra Segobin, Jérémie Pariente, Boris Suchan, Anna S. Mitchell,

Emmanuel J. Barbeau

To cite this version:

Review

The

Regulatory

Role

of

the

Human

Mediodorsal

Thalamus

Giulio

Pergola,

* Lola

Danet,

Anne-Lise

Pitel,

Giovanni

A.

Carlesimo,

Shailendra

Segobin,

Jérémie

Pariente,

Boris

Suchan,

Anna

S.

Mitchell,

* and

Emmanuel

J.

Barbeau

Thefunctionofthehumanmediodorsalthalamicnucleus(MD)hassofareluded acleardefinitionintermsofspecificcognitiveprocessesandtasks.Althoughit wasatfirstproposedtoplayaroleinlong-termmemory,asetofrecentstudies

in animals and humans has revealed a more complex, and broader, role in

severalcognitivefunctions.TheMDseemstoplayamultifacetedroleinhigher cognitivefunctionstogether withtheprefrontalcortexandothercorticaland subcorticalbrainareas.Specifically,weproposethattheMDisinvolvedinthe

regulationofcorticalnetworksespeciallywhenthemaintenanceandtemporal

extensionofpersistentactivitypatternsinthefrontallobeareasarerequired.

Nevertheless,Euryclea,takehisbedoutsidethebedchamberthathehimselfbuilt. Odyssey,BookXXIII,verses177–178.Butlertranslation.

TheMediodorsalNucleus: AReappraisal

Whenfaced withtheprospectofwelcomingastrangeras herlong-missing husband, and unabletorecognizehim,Peneloperesortedtoher‘thalamus’withthewordsreportedabove– thebedhadrootsinthefoundationofthehouseandcouldnotbemoved,adetailherhusband wouldcertainlyknow.Initiallyenteringthesceneassomethinginstrumentaltorecognition,the thalamus(seeGlossary)intheOdysseybecomesthecenterofthesceneinthelasttwobooks. Likewise,inneuroscience,theinvestigationofthemediodorsalthalamicnucleus(MD)is gainingmomentum. Untilrecently, thefunctionof theMD hasbeenmappedonto specific cognitivedomains,suchasmemoryorexecutivefunction.Aninfluentialmodelonaroleofthe MDinrecognitionmemoryforinstancesuggesteditmightplayaroleinfamiliarity[1].However, abundantevidenceindicates thatthisviewislimitedandthat theroleoftheMD inhuman cognitionmustbereconsidered.Forexample,clinicianshaveknownforalongtimethattheMD anditsbrainnetworksareinvolvedinseveralneurologicalandpsychiatricconditionsinwhich thecognitivedeficitsarenotrestrictedtomemoryfunctions[2].Neuroimagingand neurophysi-ologystudiesofthehumanMDinvivofurthersupportthisviewchange.

Thisreviewevaluatesthelatestevidenceinhumansandaimsatformulatinghypothesesto elucidatethecognitivefunctionsofthehumanMDinfuturestudies.WearguethattheMDis involvedinregulating activity patternsinthe frontallobe that arekey to perform cognitive functionscharacterizedbypersistentthalamocorticalinteractionsforlongdelays,inthefaceof

Highlights

Themediodorsalthalamicnucleusis involved in the cognitive deficits observedinseveralneurologicaland psychiatricdisorders.

Thelong-standingbeliefinaroleofthe mediodorsalthalamicnucleusmainly in long-term memory is now being reconsidered.Recentstudies empha-sizeitsfunctioninmanycognitivetasks relatedtotheprefrontalcortex. Themediodorsalthalamicnucleusis required forthe rapid and accurate performance of cognitive tasks and temporally extendsthe efficiency of corticalnetworksinvolvingthe prefron-talcortex.

Weproposethatthecommonground of multiple lines of evidence from humanstudiespointstoaroleofthe mediodorsalthalamicnucleusin reg-ulatingprefrontalactivitypatterns. Thesehypothesescanbe testedby developingspecific neuropsychologi-caltasks,parcelingthethalamuswith high-resolutionMRI,andusing intra-cranialrecordingsinhumans.

1

DepartmentofBasicMedical Sciences,NeuroscienceandSense Organs,UniversityofBariAldoMoro, Bari70124,Italy

2

ToulouseNeuroImagingCenter, UniversitédeToulouse,Inserm,UPS 31024,France

3

CHUToulousePurpan,Neurology Department,Toulouse31059,France 4

NormandieUniversity,UNICAEN, PSLResearchUniversity,EPHE, INSERM,U1077,CHUdeCaen, NeuropsychologieetImageriedela MémoireHumaine,14000Caen, France

interference,andduringmultitasking.Disruptionstothisthalamofrontalcommunicationmayin turnunderliecognitivedeficitsinseveralneurologicalandpsychiatricconditionsandrepresent apossibletherapeutictarget.

BeyondRecognitionMemory,fromRodentstoHumans

Animalmodelsemphasizing the roleofthe MD inrecognitionmemory andfamiliaritywere basedonitsmonosynapticinputfromtheperirhinalcortexinprimates[1](whilethispathwayis only weak in rodents [3]). Earlier pioneering work in non-human primates, however, had demonstratedtheinfluenceofMD–prefrontalcortex(PFC)interactionsonMDactivityacross delays[4,5].Recentevidenceinrodents[6–8]andinmonkeys[9,10]indicatesthattheMD influencesmultiplecognitiveabilitiesviaitsinteractionswithareasofthefrontallobe,suchas thePFCandanteriorcingulatecortex,towhichtheMDisreciprocallyconnectedinrodentsas wellasinprimates[11–13](Figure1).IndividualMDneuronsfromdifferentsubdivisionsofthe MDexhibitaconsiderabledegreeofdivergenceintheirprojections,thatis,eachrodentMD neuronprojectstoseveraldifferentPFCsubdivisions[14].Similarly,inmonkeysandhumans, MDefferencesdivergeandmakecontactwithmultiplePFC areas[15,16].TheMDis thus interactingwithmanyfrontalareassimultaneously,whichinturnhaveintrinsicconnections amongmultiplecorticallayers[17].MDneuronsmaydirectlyfacilitatecorticocortical commu-nicationviatrans-thalamicpathways[18],andit wouldbeimportant tofurthersupportthis hypothesiswithneurophysiologicalmeasurements[19].

WhathasbeenconvincinglydemonstratedisthatpersistentPFCactivitypatternsdependonMD inputsandonrecurrentexcitationofthalamofrontalcircuits[6,20].HencetheMDmayhavearole, notlimitedtolong-termmemory(LTM),insustainingdelay-relatedactivityinthePFC[21]. Interestingly,whiletheinitialmaintenancemaybesustainedbythePFCalone,itsinteractionswith theMDcouldextendthisactivitypatternfromseveralsecondstoseveralminutesandpossibly beyond[21].ThetemporalregulationofthemutualinterdependenceofMDandPFCactivity,for example,therapidadjustmentofthephaseandfrequencyofcorticaloscillations[7,9,13],isan ideaalsocommontootherviews[18,22,23].Thus,theMDwhenactivelyengagingwiththePFC mightsupportsynapticreverberationsinrecurrentthalamofrontalloopsthatpromotepersistent activityacrossseveralcorticalregionsnecessaryforefficientcognitivefunctioning.Inotherwords, theinfluenceoftheMDonthecortexmayallowforreflections,decisions,andactionsrelevantto thecurrenttaskdemandstoextendoverawindowoftimethatiscontextuallyrelevantandunfolds attemporalscalesdistinctindifferentmammalianspecies.

Theseideas,mainlydevelopedthroughexperimentsinrodentsandmonkeys,areplausiblein humanstoo,andtheseadvancescallfortimelytranslationsintothehumanfield.Wearewell awarethatthereisstillmuchtolearnintermsofestablishingclearhomologiesbetweenanimal modelsandhumans.Forexample,theprimateMDincludesanintrinsicpopulationof inter-neuronsreleasingGABAthathasnotbeenidentifiedinrodents[24].Inprimates,theMDisalso richindopaminereceptorsreceivingtheirinputfrommultipleindependentpathways[25,26], makingitpartofwell-studiedbrainnetworksinvolvedinsaliencydetection[27].Further,theMD ispartofaprimate-specificnetworklinkingtheamygdalawiththethalamicreticularnucleus

[28],oneofthemainsourcesofGABAwithinthethalamus.Importantly,animalsaretypically overtrainedonthetaskstheyperform,whereasithasbeenarguedthatnovel,complextasks, noteasilysolvedbasedonprocedures,expertise,oroverlearnedknowledge,should particu-larly tap MD–PFC interactions [13]. Yet we contend that general principles learned from experimentalanimalmodelsarenotunderminedbythesedifferences,becauseofthegenerally

5

DepartmentofSystemsMedicine, TorVergataUniversityandS.Lucia Foundation,Rome,Italy

6

ClinicalNeuropsychology,Ruhr UniversityBochum,

Universitätsstrasse150,44801 Bochum,Germany

7

DepartmentofExperimental Psychology,UniversityofOxford,The TinsleyBuilding,MansfieldRoad, OxfordOX13SR,UK

8

CentrederechercheCerveauet Cognition,UMR5549,Universitéde Toulouse–CNRS,Toulouse31000, France

9

Equivalentcontributionaslast authors

*Correspondence:

giulio.pergola@uniba.it(G.Pergola)and

anna.mitchell@psy.ox.ac.uk

Glossary

Attentioncontrol:attention-related tasksinrealliferequireignoringa varietyofdistractionsandinhibiting attentionshiftstoirrelevantactivities. Attentioncontrolconsistsofthetop– downallocationofattentional resourcestoperformavarietyof cognitivetasks.

Executivefunctions:thecapacity ofthebraintoformulategoals,plan, andcarryoutplanseffectively[98]. Long-termmemory:theabilityto rememberlearnedmaterialfortime spansfromminutestoalifetime, includingavarietyofmemory systems(e.g.,episodic,semantic, procedural),whichsharethestorage ofmemoryrepresentationsforatime exceedingthepersistenceofthe informationinthestreamof consciousness.

Mediodorsalthalamicnucleus: nucleusofthedorsalthalamus, locatedintheanteroposterioraxis belowtheanteriornucleus,atthe midline,mediallytotheinternal medullarylamina.Ithasatleasttwo differentsubdivisions:themedial magnocellularMDandthecentral parvocellularMD.Athirdsubdivision, thelateralMD,isincludedbysome authorsamongtheintralaminar nuclei[94].Thesesubdivisionseach receiveafferentsoriginatingfrom differentpartsofthebrainstem, midbrain,basalganglia,prefrontal cortex,andlimbicsystem.Different MDsubdivisionsprojecttodifferent frontalareas,andanteriorcingulate andinsularcortex.Theparamedian andtuberothalamicarteriessupport perfusiononitsmedialand rostrolateralborders,respectively. MRI,structuralandfunctional (fMRI):Structuralimagingprovides staticanatomicalinformation translatingthelocalmolecular differencesintodifferentshadesof graytooutlinetheshapeandsizeof thebrainregions.AnMRIscanner deliversaspecificradiofrequency thatexciteshydrogenatoms,which returnsomeofthisenergyinthe formofacharacteristicnuclear magneticresonancesignal. Functionalimagingsuppliesdynamic physiologicalinformationindirectly relatedtometabolicchangesinthe neuraltissue,includingbloodoxygen level-dependentcontrast,perfusion (whetherbyendogenousor similarconnectivitypatternsoftheMDwiththePFCacrossspecies.Ifanything,the

primate-specificMDfeaturesmakeitmorecentralinbrainnetworksrelevanttocognition(Figure1). Herewepropose that suchspecies-specificadaptations,ratherthan establishing different functionsintheprimatecontextcomparedtorodents,reflectthephylogeneticadaptationofthe interactionsbetweentheMDandthePFCinthecontextofagenerallydifferentorganizationof theprimatebrain. Forexample,whilesaliencydetectionandpersistentPFCactivity across delaysareaffectedbyMDdysfunctioninrodents,theneurochemicalbasisislikelydifferentto thatinprimates,sincethelattertakesadvantageofdopamineinputstotheMD.Specifically,we arguethatthehumanMDisinauniquepositiontoparticipateintheactivityofmultiplebrain networksthatexceedthedefinitionofasinglecognitivedomain–anotionthat,inrodentsand non-humanprimates,issupportedbymultiplelinesofevidence[7,13,29].Tothis aim,we reviewtheavailableevidenceinhumanstudiesfromtheperspectiveofclinical,neuroimaging, andneurophysiologicalstudiesthathighlighttheimportanceofMD–PFCinteractions.

ThalamicStrokeStudies

Inhumans,theMDhasinitiallybeenassociatedwithLTMbasedonitsassumedinvolvementin Korsakoff’ssyndrome(KS)[30,31](Box1reportsahistoricalperspectiveofthefunctionofthe MD inmemory). Thalamicstroke studies have also historicallyplayed an important role in shedding light on the function of individual thalamic nuclei. In particular, ischemia in the paramedianortuberothalamicarteryorhemorrhagecauses MDdamage [32,33].Because ofthesmallsizeofthalamicnuclei (Box2),vascularlesionsarenecessarilyunselectiveand involvemultiplenuclei,whichalsoplayaroleintheensuingcognitivedeficits.Forthisreason,it iscrucialtoquantitativelyestimatethevolumelossseparatelyfordifferentnuclei,anapproach thatbecameviableonlyrecentlythankstoadvancesinneuroimagingtechniquesbutthathas beentoorarelyundertakensofar[34–36].Lesionquantificationischallengingespeciallyfor lesionsproximaltothethirdventricle:theseinfarctstendtomergewiththeventricle,orthe ventricleitselfundergoesprogressiveenlargementassociatedwithtissueshrinkage,hindering volumemeasurementsinthemedialnuclei(seeOutstandingQuestions).

Despitethesemethodologicallimitations,twoofthelargestgroupstudiesofischemicfocal thalamiclesionstodateagreedonamild-to-moderateLTMimpairmentofchronicpatientswith MDlesions,whichcouldnotbeexplainedbyconcurrentlesionsofthehippocampal–thalamic axis[36,37].Short-termmemory,includingworkingmemory(WM),deficitsarenot consis-tentlyreportedingroupstudies,withfewpositivefindings[36,38].Thislackofevidenceabout WMdeficitsinstrokepatientswithfocalMDlesionsisimportantbecausereportsbasedon some animal models emphasized a role of the MD in WM [39–41]. The poor consensus betweenclinical reportsmightalso berelated tosparsehumanevidencefollowingbilateral lesions.Thesebilaterallesionslikelycausemoresevereimpairmentthanunilateralones,but occurmorerarely.Studiesinpatientswithbilaterallesionsmaythusrevealdeficitsotherwise toomildtobeclearlyidentifiedinpatientswithunilaterallesions[33,42].Othermethodological issuescouldalsocontributetotheconflictingevidence:forexample,theuseofspantasks reflectsmoredirectlypassivestorageabilitiesthanothercomponentsofWM(i.e.,manipulation, interferencecontrol,orupdating)[43].Basedonfurtherfindingsreviewedbelow,thesekey skillsmaybeparticularlyaffectedafterMDdamage.

Executive functions, attention control, prospective memory, arousal, motivation, lan-guage,andbehavioraldeficitsarealsooftenreportedintheacutephaseoffocalMDlesions

slightmemoryproblems.Thefunctionaloutcomeofthesepatientsislargelyunknown(see OutstandingQuestions).Onthisbasis,thereisnoagreementonaclinicallyrelevantchronic outcomeofMDdamage[46],exceptperhapsforamildLTMimpairment.

Overall,thelossofcognitivefunctionsafterMDdamageinhumansappearspoorlydefined.Itis possiblethatdamagetotheMDisneithernecessarynorsufficienttoinstantiatechronicdeficitsin othercognitivedomainsthanLTM. Alternatively,andthisistheoptionweexplorehere,the standardtestsusedtorevealnon-mnemonicdeficitsmaybeinsufficientlysensitiveforelucidating thekindofimpairmentsthatoccurinhumansafterMDlesions.Forexample,manypatientswith frontallobe damageshow littledeficitson standardtests,yetare severelyimpaired intheir professionalandfamilylives[47].Inthe1990s,thislackofclearimpairmentinlaboratorytests was rectifiedwith the developmentof novel tests (i.e.,requiringperformance ofseveral tasks within alimitedamountoftimeusingastrategythatthepatientshavetodevelopthemselves;Box3).We proposethat,justaswasthecasewithfrontallobedysfunctionsbackinthe1990s,deficitsgoing beyondLTMimpairmentshavelikelybeenunderestimatedbecauseofthepaucityofcases,tests employed,andconfoundingeffectsoflesionlateralityalongwithpoormeasurementsoflesions. Further,wesuggestthatthedevelopmentofadhocneuropsychologicalteststoinvestigatethe MD may reveal novel insight especially regarding the temporal parameters affecting task performance, such as reaction times and response deadlines. Animal studies revealed a functionoftheMDinpersistentPFCactivitywhichwouldbeconsistentwiththeimpairments shown by patients with frontal lobe damage in self-paced executive tests. However, the availableevidenceonthetemporalparametersinpatientswithMDlesionissparseandthe neuropsychologyfindingsonthistopicarecontradictory(e.g.,[48]reportednoeffects;[37]

reportedincreasedreactiontimeinpatientsrelativetohealthycontrols). ThisaspectofMD functionmay even affect theattribution ofdeficits to underlyingprocesses basedontheir timescale.Forexample,familiarity-basedresponses(Box1)maybefasterthan recollection-basedones [49,50]. Notably, ifthe different temporal regulationof cognitive operationsis collineartotheoperationsathand,thenwhatappearstobeaqualitativedifferencebetween separate cognitive substrates may in part be related to an underlying role of the MD in supportingtemporalaspectsoftheperformance.Asanotherexample,onalongertimescale (24hafterlearning),patientswiththalamicischemiaencompassingtheMDandotherthalamic nucleishowacceleratedforgetting[51].ThisiscaseinpointthatLTMdeficitsofpatientswith thalamiclesionsmaybepartlyexplainedbyaroleoftheMDforexampletemporallyextending– highercognitivefunctionssubservedbyfrontallobeareas.

ClinicalConditionswithGradualDevelopmentofMDDysfunction

AmonganumberofpathologiesthathavebeenassociatedwiththeMD,alcoholusedisorder (AUD; see Box 4), KS, and schizophrenia (SCZ) are of particular interest. The changes associatedwiththesedisordersusuallyevolveslowlyovertime.Thus,thisevidence comple-ments the lesion studies as it reflects gradual rather than abrupt loss of function. This characteristicyieldsthepotential tofollow therelationshipbetweenneuroimagingreadouts andcognitive/behavioraloutcomesovertime.

TheMDandotherregionsofthemedialdiencephalonhavebeenproposedtounderliethe pathophysiologyofKS[52].Invivoneuroimagingstudieshaveshownshrinkageofanteriorand medialthalamicnucleiinpatientswithAUDandKS([53];seeBox4).Consistently,AUD,KS,as wellasSCZarecharacterizedbydeficitsofattention,WM,andexecutivefunction [54,55]. However, jointevaluationsof the neurologic andpsychiatric literatureaimed to inform the investigationofMDfunctionshavebeenraresofar[13,56].

exogenouscontrast),bloodflow,and cerebrospinalfluidpulsation. Prefrontalcortex:themostrostral partofthefrontallobe,which coordinatesawiderangeofneural processesandincludes

interconnectedneocorticalareasthat sendandreceiveprojectionsfrom virtuallyallcorticalsensorysystems, motorsystems,andmany subcorticalstructures. Thalamus:acomplexof50–60 nuclei,locatedinthediencephalon, spanningbothhemispheres.The rostro-caudaldimensionofthe humanthalamusisabout30mm,its heightabout20mm,anditswidth about20mm,withabout10million thalamicneuronsineach hemisphere.

InSCZ,post-mortemstudiesrevealedgraymatterreductionandneuronallossinthethalamus ofpatients,althoughthisevidenceisunspecificwithrespecttotheMD[57,58].Neuroimaging studiessupporttheideaofthalamicneuropathologyinpatientswithSCZ[59],withlongitudinal graymatterchangesinthethalamusassociatedwithcognitionmeasures[60].Volume loss appearsnonhomogeneous acrossthalamicnuclei andshowsgreatereffects inthe medial aspectsofthethalamus[61].Unfortunately,veryfewneuroimagingstudiesofpatientshave performedthalamic parcellation,andneuroimaging quantitative assessmentsofthe medial thalamusmay beconfoundedby enlargements ofthe thirdventricle. Nevertheless, recent

MDpc

Rostral

pallidum

Dorsal

caudate

SNr

Amygdala

VMPFC

/ OFC

Entorhinal,

perirhinal,

parahippocampus

DLPFC

BA 6

BA 8

Ventral

pallidum

Ventral

striatum

SNr

VTA

Midbrain

Brainstem

TRN

DLPFC, VLPFC

DMPFC, VMPFC, OFC

MDmc

MDpc

MDpc

PFC

ACC

MDmc

Hippocampus

Box1.FamiliarityorRecollection?AnHistoricalOverview

Theinterestonthalamicnucleiashigher-ordercognitionsubstratesenjoyedawidespreadincreaseafterAggletonand Brown’sreview[1].Theauthorsdescribedthesubstratesofthetwoprocessesunderpinningrecognitionmemory: recollection,theabilitytoretrievepartoftheexperienceassociatedwithastimulus,andfamiliarity,themerefeelingthata stimulushasbeenexperienced.Theysuggestedthatthecircuitlinkingthehippocampuswiththeanteriorthalamic nuclei(ATN)alongwiththemammillarybodiesandthemammillothalamictract(MTT)supportedrecollection.Theyalso proposedthatasecondindependentcircuitinvolvedtheperirhinalcortexandtheMDprocessedfamiliarityduetotheir directconnections.AlthoughacriticalrolefortheATNinrecollectionremainsundisputed,theroleoftheMDinfamiliarity isstillcontested.Indeed,studieshavetypicallyreportedimpairedrecollectionwithrelativelypreservedfamiliarity followingMDdamage[38,99–101].Aggletonandcolleagues[102]thusrevisedtheirmodelintegratingthespecific connectivitypatternofeachthalamicnucleus.Themultieffectmultinucleimodeldescribedafunctionalcontinuum, ratherthanadissociation,betweentheMTT/ATNandMDviathemidlineandintralaminarnuclei.Inparticular,they proposedrecollectiontobeimpairedfollowingMDdamagebecauseofthedenseconnectionsbetweenthisnucleus andprefrontalareas,henceswitchingtheroleofthisnucleusawayfromitsrelationwiththemedialtemporallobes.A distinctionwasalsodrawnbetweentheparvocellularMD,whichmaybeinvolvedinrecollection(duetotheirdense connectivitywiththePFC),andthemagnocellularMD,whoseroleremainsmoreelusive[102].Subsequentstudies, includingthosewithmorerefinedimagingapproachestolocalizelesions,appearedinagreementwiththeseproposals

[36,37,51],leavingthepurportedroleoftheMDinfamiliarityunsubstantiatedwiththeexceptionofasinglecasestudy

[103]andfMRIstudies[74–76].Recently,asinglecasestudyassessedtheimpactofMDdamagesustainedatbirthon theForcedChoiceCorrespondingtest,whichrequiressubjectstorecognizestimuliamongsimilarfoils.Performanceon thistaskisthoughttocriticallydependontheperirhinalcortex,and,byextension,ontheMDasatrans-thalamicrelayto prefrontalareas.Thepatientwasindeedimpairedontrialsfilledwithvisualinterference,whichindicatesthattheMDmay beinvolvedinsomevisuallydemandingrecognitionmemorytasks,withoutnecessarilymappingontheclassical familiarity/recollectiondistinction[42].

Box2.TrendsinMagneticResonanceImagingofThalamicNuclei

FromanMRIperspective,measurementsneedprecision(theminimumpossibleerrorinestimatingthesignalinavoxel) andaccuracy(freeofartifacts,well-localizedsignals).Owingtotheirsmallsize[104]andsimilarityintermsofrelaxation timesand/orprotondensities,segmentingthalamicnucleiisparticularlychallenging.

StructuralScans

Manualsegmentationofthalamicnucleidependsongoodcontrastbetweennucleiofinterestandneighboringregions. Theoptimumchoiceistousehigherfieldstrength(e.g.,7T),offeringhighercontrastandsignal-to-noiseratio(SNR). Acquisitionsaretypicallystudyandnucleispecific.Priorreportsemployedinversionrecovery-turbospinecho[105]for imagingthedorsalthalamus,susceptibility-weightedimaging[106]foritsventralintermediateaspects,andMPRAGE sequencesthatnullifywhitematterthatseparatesseveralnuclei[107].Atrade-offisneededbetweenSNRand acquisitiontimes.Shorteracquisitiontimeshelpminimizemovementartifactsduetoheadmotionorcerebrovascular pulsation[105],whilelongeracquisitiontimesallowforhigher-resolutionimages,decreasingthemixtureoftissuesina voxel.Typically,7-TMRIsequenceswithdiscernablenucleihadacquisitiontimesof7–15minforimageresolutions varyingfrom0.67mmisotropicto0.3750.3751mm3

[107].

Currently,withmostdatacollectedatlowerfieldstrength(3T),automaticsegmentationisoftenpreferred.Itcanbe achievedthroughahistologicalatlas[108]andrequiresnormalizationoftheMRIintoatlas-standardizedspace.Its accuracyislimitedtotheresolutionofboth,MRIandatlas,anddoesnotaccountforintersubjectvariabilityofshapeand volumeforeachnucleus,especiallyinpatients[109].

StructuralConnectivity

Automaticsegmentationcanalsoemploydiffusionimagingandstate-of-the-arttractographyalgorithms.The con-nectivitystrengthineachthalamicvoxelisevaluatedwithrespecttoapriori-definedregions[110–112]oreveryvoxelin thebrain[113]andthenclusteredtogetheraccordingtoconnectivity-basedfeaturesimilaritiestosegmentthethalamus

[114].

FunctionalConnectivity

Resting-statefMRIstudieshavesegmentedthethalamusbasedonfunctionalconnectivitypatternswithcorticalareas (e.g.,independentcomponentanalysis[115,116]ornormalizedspectralclustering[117]).Thalamicparcelstypicallydo nothaveaone-to-onemappingtocorticalregionsandaresharedamongfunctionalnetworks[113].

Box4.TheThalamusinAlcoholUseDisorder,Korsakoff’sSyndrome,andDiet

Historically,thelinkbetweenthethalamusandcognitionoriginatesfromstudiesonKS,primarilywithexcessivealcohol consumption[125].Alcoholismmainlyaffectsthefronto-cerebellar(includingtheMD)andPapezcircuits[126],which sharethethalamusasakeynode.Recentneuroimaginginvestigationshaveconfirmedneuropathologicalstudies, detailingalterationstothalamicvolumeandstructuralconnectivityinAUDpatientsevenwithoutKS[53,127].KSonsets whenexcessivealcoholconsumptioniscombinedwiththiamine(vitaminB1)deficiency(TD),andischaracterizedbya profound,globalamnesia.AUDpatientsareatspecialriskforTDnotablybecauseofalteredthiaminemetabolism.It remainsunclearwhetherthemarkedchangestothebrainobservedinKSoccurasaresultoftheneurotoxiceffectsof alcohol,orsustainedTD,oracombinationofboth[128].Thebrainandneuropsychologicalrecoveryobservedafter abstinenceinpatientswithAUDwithoutamnesia[54,129]suggeststhatalcoholismalonemaynotsystematicallyleadto persistentbraindysfunction.GlobalamnesiainKSremainsevenaftercessationofalcoholuse.Thus,severeand persistentdamagetothethalamusobservedinKSlikelyresultsfromTDratherthanalcoholperse,asalsosuggestedby thedescriptionofKSwithoutahistoryofAUDbutwithsystematicnutritionaldeficits(e.g.,bariatricsurgery,anorexia). AnimalmodelshavebeenessentialtodeterminetherespectivecontributionsofexcessivealcoholconsumptionandTD tothedevelopmentofalcohol-relatedbraindamage[130].Thesecausativestudiesinrodentshaveestablishedthat chronicandheavyalcoholintakeisnotmandatorytomimicthespecificthalamicalterationsobservedinKS[131],but alcoholmaypotentiatetheeffectsofTD[132].InAUDpatients,alteredthiaminemetabolismwassolelypredictiveof episodicmemoryimpairments[133]andlowerlevelsofcirculatingthiaminediphosphateselectivelycorrelatedwith poorerepisodicmemoryperformance[134].

ThethalamusisnothomogeneouslyaffectedbyTD.Themedialandmidlinethalamicnuclei,andtheanteriorthalamic nucleiareespeciallydamagedinKScomparedwithAUDpatientswithoutamnesia[53],andinpyrithiamine-inducedTD rats[130],reinforcingtheideathatthesenucleiandtheirconnectionsplayacrucialroleinmemory[135].Bycontrast, theMDisdamagedinAUDpatients,butnotespeciallyinKSpatients,orinanimalmodelsofKS.Thus,the fronto-cerebellarcircuit,includingtheMD,maynotbeespeciallyvulnerabletoTD,butrathertoothercomorbidalcohol-related braindysfunction[133].

Box3.CognitiveTaskstoAssessMDFunctionsinHumans

Wesuggest,inthisreview,thatmoststandardneuropsychologicaltestsarerelativelyinsensitivetoidentifyinghuman MDfunctioning.TheintrinsicconnectivityofthePFCclearlysupportsmanycognitivefunctionsonitsownandtherefore humanMDlesionsmaycauseonlymoderateornonspecificimpairmentsinPFCexecutivefunctionswhenassessed withstandardneuropsychologicaltests.Inthiscontext,specialtestsareneededtoaccountforthespecificcontribution oftheMDtocorticalPFCfunctioning.GiventhatMDneuronsareinterconnectedtomanyPFCregions,theirroleis probablymoreevidentintaskswithmultifacetedcognitivedemands.

WeproposetheuseofteststhatmeetsomeofthefunctionalcharacteristicsoftheMDoutlinedinthisreview.For example,manipulationofinternalrepresentations,includingmemory,predictivecoding,goal,rules,susceptibleto degradationduetocognitiveload,adaptivedecisionmaking,multitasking,interference,orlongdelays,uptoseveral minutes(e.g.,>5–30min)requiresstronginteractionsamongPFCregionsaswellastemporalandspatialextensions, andhencemayidentifyanMDcontribution.

evidenceidentifiedMDgraymatterestimatesasthetop-rankingthalamicfeaturediscriminating patientswithSCZfromcontrolsusingmultivariatestatisticalanalyses[62].Likewise, longitu-dinalchangesinthalamicgraymatterinpatientswithSCZappearlocalizedespeciallyinthe midlinethalamicnucleiandMD[63].

The discrepancy between the post-mortem and structural neuroimaging evidence invites caution,as motion artifacts, effects ofmedicationon brainperfusion, andmetabolic state maybiasgraymatterestimates[64].Inaddition,itisunclearwhetheralterationsintheMDarea causeoraconsequence ofthe disease,as multiplestudies failedto associatedecreased thalamicgraymatterwithgeneticriskforSCZ[61].Therefore,MDdamagemaybea conse-quence of the illness course, potentially confoundedby medication or symptom severity progression[58].

Relevanttothisreview,functionalimagingtasksrevealdifferencesbetweenpatientswithSCZ andhealthycontrolsthatarenotconfinedtoLTM,althoughepisodicmemoryalterationsare well supported [65]. For instance, medial thalamic regions are hypoactive in SCZ during attentionandWM tasks[61].Inaddition,thefunctionalconnectivity betweenthethalamus andthePFCisdecreasedinpatientswithSCZ,andintheirsiblings,bothduringrestingstate

[66–68]andduringattentioncontrol[69].Thisthalamo–PFCfunctionalconnectivityalterationat restingstatewasalsofoundinindividualsatriskorinearlydiseasestagesandwasassociated withverballearningandmemoryperformanceinpatientswithpsychosis[70,71].Thethalamic regiondisconnectedfromthePFCwaslocatedinamedialthalamicterritorycompatiblewith the localization of the MD [69,70]. Insummary, althoughonly few studies considered the heterogeneityofthalamicnuclei,theMDanditsPFCnetworksappeardysfunctionalinpatients withSCZandintheirrelatives,associatingfunctionalchangesinMDactivityandconnectivity withthe genetic component of SCZ, with illnesscourse, and importantly, witheffects on cognitionwiderthanLTM.Itwillberelevantforfuturestudiestocharacterizethecognitiveand clinicalcorrelatesofsuchalterations,asthelinkofMDdysfunctionwithlongitudinalaspectsof theillnesshighlightsthatsuchMD–PFCinteractionsmayrepresentatherapeutictarget(see OutstandingQuestions)[60,63,71,72].

NeuroimagingandNeurophysiologyRecordingsoftheMD

IftheMD isnotdirectlyrelated toaspecificmemorycomponent,itmay rathersubservea generalrole ingoal-directed behaviorbeyond LTM,that is,inpersistentactivity underlying differenttypesoflearning[77–82].SeveralstudieshavefurthersuggestedthattheMDmay processthe allocationofattention andtheinteraction betweenattentionand learning pro-cessesinatask-relevantway[29,83].Inthislight,thesignaldetectedintheMDduringepisodic memoryperformancemayrepresentthetemporalactivationofarecurrentfronto-thalamicloop thatisbeingmaintainedduringinformationprocessingincorticalnetworks.Itfollowsthatthe humanMDmaybeactivatedwhentasksrequirethemaintenanceofpersistentneuralactivityin areasofthefrontallobeandbeyond.

Intracranialneurophysiological recordings in the human MD provide criticalinsight for this proposal.Asingle-patientstudyfoundthatstimulus-linkedoscillatorysynchronybetweenthe MDandfrontalsurfaceelectrodeswasenhancedforsuccessfulrecognitionmemoryretrieval comparedwithsuccessfulcorrectrejectionsofnewitems[84].AGrangercausalityanalysis suggestedthatthedirectionofthisconnectivitywasthalamocortical,hencesupportingtheidea that the MD would enhance prefrontal activity during LTM retrieval. Another intracranial neurophysiologicalstudyassessingbothencodingandretrievalfoundthatMD prestimulus activityduringencodingpredictedmemorysuccessinanincidentalencodingtask[85].MD synchronywithfrontalthetawavespredictedsuccessfulencoding,consistentwithfMRIand lesionevidenceontheinvolvementoftheMDduringencoding[56,73].Inaddition,evenMD restingstateactivityunrelatedtothetaskwasassociatedwithsuccessfulmemoryformation acrossparticipants.TheseresultssuggestthatMD–PFCinteractionsareassociatedwithan overallcognitivedispositiontosuccessfulmemoryformation,evenwhenthatisnotthetask goal–whichissurprisinginlightofthefMRIliteraturesupportingaroleoftheMDingoal-related behavior,and hence requires specialconsiderationin models ofhumanMD function. The findingsfromathirdstudyfurthererodedtheconceptthattheMDisprimarilyinvolvedinLTM

[86].Agroupofpatientswithepilepsyundergoingintracranialelectrodesurgeryfordeepbrain stimulation performeda complex executive function task tappinginto attention, WM, and decisionmaking.ReversibleMDdysfunctionobtainedbyapplyinghigh-frequencystimulation causedsignificant deficits inthe task.TheauthorsconcludedthattheMD connects retro-spectivesensorywithprospectiveactionrepresentations.Onthewhole,intracranialrecordings suggestaroleoftheMDindirectingcorticalallocationofattentionandtherebysettingthestage forpersistentcorticalactivitytooccurbyregulatingprefrontaloscillationsinatime-sensitive manner.

TheMD: AnEnhancerofFrontalLobeFunction?

Influentialviewsonthefunctionsofthethalamushaveemphasizeditsactivegatingproperties withrespecttostimulidirectedfromtheperipherytothecortex,orfromthecortextoother corticalareasviatrans-thalamicroutes[87–90].ThehumanPFCmaybeabletoundertake manytaskswithoutafunctionalMD,evenWMtasksthat,inrodents,arechallengedbyMD lesions. However, the evidencewe reviewedsuggeststhat the MD may actively enhance prefrontalexcitability(i.e.,increasetheamplitudeordurationofcorticalactivity)[91].Active enhancingallowsamorenuancedinfluencefromtheMDoncorticalfunctioningcomparedwith gating.SuchamodelisconsistentwiththeideathattheeffectofMDdysfunctionmayonly becomeapparentwithlongerdelaysordemandingtasksthataretemporallyextendedbeyond thereachofWM(perhapswitharoleinpromotingprefrontalplasticity[56]).Nevertheless,some questionsmightremainunansweredbythe‘enhancer’model.SincethePFCalreadyhosts reverberating circuits, research on the human thalamus needs to investigate the specific contributionofMD-mediatedcorticalprocessing.

TheTrans-Thalamic Route:TheMDasaRegulator

Inhumans,recurrentcircuitswithinthePFCmaybesufficientforshorttimeintervals(e.g., withinthespanofWM),whiletheMDregulationofprefrontaloscillationsmaypromotethe temporal extension of PFC activity. A recent study [92] notedthat the precision of WM representationsdegradesacross longdelays andfurther mechanismsmaybeneeded to preserveitforlongerretentionintervals,mechanismsbeyondthe‘enhancement’ofspiking activity.For example, the currently prevailing WMmodel includes a dedicated system to preservemultimodalmemoryrepresentationsacrossdelays(episodicbuffer)[93].Besidesits enhancingcapacity,intracranialrecordingsshowthattheMDregulatescorticaloscillations throughsignalsdirectedfromtheMDtothePFCwithabehavioralsignificance–evenbefore stimulusonsetandevenunrelatedtoexplicitlydefinedtaskgoals[85].TheMDmaythusbe part of a network bridging past with future activity patterns across multiple cortical PFC regions[18].This‘connecting’rolemayalsoexplainwhyitsfunctionhasbeenelusivesofar: mostoftheoperationsareperformedinthePFC.Notably,aregulatingroleoftheMDdoes notnecessarilyimplythatitdrivescorticalactivity,butrathersupportsitduringlongerepochs oftime.During thislapse oftime,dynamicshiftsbetween theseregionsmayberequired dependingontheir functional specializations,which mightrepresentinterferingprocesses thatultimatelyaffectrepresentationprecision.

persistentactivitypatterns,and(ii)whetherornottheinformationisupdated,byantagonizing representationdegradation.ThefactthattheMD,likethePFC,projectstomultiplefieldswithin thethalamicreticularnucleus(whereasothernucleiareconnectedtospecificthalamicreticular fields[97])alsoenablestheMDtoinfluenceotherthalamocorticalnetworks.Thiscircuitextends itsinfluenceacrossotherregionsofthebrain,hencealsosuggestingarolefortheMDinthe spatialextensionofthesignal.

Thismodelcould explain why therole of theMD may be mostrelevant when the PFCis multitaskingforlongtimeintervals.TheseaspectsofMDfunctionsmaybebestinvestigated usingcarefullycontrolled interference[42] instudies affordinghigh temporalresolution,for example,intracranialrecordings(seeOutstandingQuestions).Overall,understandingtherole

KeyFigure

Regulation

of

the

Prefrontal

Cortex

MD mc MDpc VMPFC DLPFC frontalOther areas Amygdala Perirhinal cortex Striatum VTA / SN TRN Other corƟcal areas Other thalamic nuclei

Fron

tal lobe

oftheMDinhumancognitionwilllikelyrequirededicatedteststappingintovariouscognitive functionsratherthanLTMalone,withafocusontemporalparametersofthetask(Box3).

ConcludingRemarks

LikeUlysses’bedintheOdyssey,thethalamusisrootedinthecenterofthebrainandthis featurehas constituteda formidablechallenge for humanneuroscience research. Asnew techniquesallowustopeekinsidethefunctionsofspecificthalamicnuclei,theMDisemerging fromitspurportedfunctioninsupportingrecognitionandis,instead,beginningtocommanda higherprofileincognitive,behavioral,andclinicalneuroscience.

WesuggestreconsideringthefocusonLTMthathascharacterizedpartoftheliteratureinthe pastyears.WearguethattheMDroleismorewidelyrelatedtothemaintenanceandtemporal extensionofpersistentactivityinthefrontallobes.Newstudiesshouldinvestigatethalamic nuclei separately with multimodal imaging assessments whenever possible (with special considerationoffunctionalconnectivityapproaches),andincludelesionquantification,possibly accountingforbiasintheneuroimagingestimates.Onlyspecificallydesigned neuropsycho-logical tests– as opposedto routineassessments– intandem withstate-of-the-art fMRI sequencesanddataanalysisperformedinaconsortiumframeworkwillachievesamplesizes suitabletoapproachtheconundrumonthefunctionofthehumanMD.

Acknowledgments

Thisreviewwasinspiredfromdiscussionsamongtheauthors,severalofwhompresentedatthethalamicsymposium ‘What’sthischamberdoinginmybrain?Theroleofthethalamusinmemory’attheInternationalConferenceonMemory (ICOM6)inBudapest,July2016.G.P.hasreceivedatravelawardforanacademicexchangeprogramfromthenon-profit organizationBoehringerIngelheimFonds,leadingtothiswork.A.L.P.isfundedbytheInstitutUniversitairedeFrance.B.S. wasfundedbyagrant(Sonderforschungsbereich874,CRC874)fromtheGermanResearchFoundation(Deutsche Forschungsgemeinschaft,DFG,ProjectB8).A.S.M.issupportedbyaWellcomeTrustSeniorResearchFellowshipin BasicBiomedicalSciences110157/Z/15/Z.

References

1. Aggleton,J.P.andBrown,M.W.(1999) Episodicmemory, amnesia,andthehippocampal-anteriorthalamicaxis.Behav. BrainSci.22,425–444

2. Golden,E.C.etal.(2016)Mediodorsalnucleusanditsmultiple cognitivefunctions.Neurology87,2161–2168

3. Furtak,S.C.et al.(2007) Functionalneuroanatomy ofthe parahippocampalregionintherat:theperirhinalandpostrhinal cortices.Hippocampus17,709–722

4. Alexander,G.E.andFuster,J.M.(1973)Effects ofcooling prefrontalcortexoncellfiringinthenucleusmedialisdorsalis. BrainRes.61,93–105

5. Fuster,J.M.andAlexander,G.E.(1971)Neuronactivityrelated toshort-termmemory.Science173,652–654

6. Bolkan,S.S.etal.(2017)Thalamicprojectionssustainprefrontal activityduringworkingmemorymaintenance.Nat.Neurosci.20, 987–996

7. Schmitt,L.I.etal.(2017)Thalamicamplification ofcortical connectivitysustainsattentionalcontrol.Nature545,219–223 8. Alcaraz,F.etal.(2018)Thalamocorticalandcorticothalamic pathwaysdifferentiallycontributetogoal-directedbehaviorsin therat.eLife7,e32517

9. Browning,P.G.etal.(2015)Evidenceformediodorsalthalamus andprefrontalcortexinteractionsduringcognitioninmacaques. Cereb.Cortex25,4519–4534

10. Chakraborty,S.etal.(2016)Criticalroleforthemediodorsal thalamusinpermittingrapidreward-guidedupdatingin stochas-ticrewardenvironments.eLife5,e13588

11. Dermon,C.R.andBarbas,H.(1994)Contralateralthalamic projectionspredominantlyreachtransitionalcorticesinthe rhe-susmonkey.J.Comp.Neurol.344,508–531

12. Xiao,D.etal.(2009)Laminarandmodularorganizationof prefrontalprojectionstomultiplethalamicnuclei.Neuroscience 161,1067–1081

13. Ouhaz,Z.etal.(2018)Cognitivefunctionsand neurodevelop-mentaldisordersinvolvingtheprefrontalcortexandmediodorsal thalamus.Front.Neurosci.12,33

14. Kuramoto,E.et al.(2017)Individualmediodorsal thalamic neuronsprojecttomultipleareasoftheratprefrontalcortex: asingleneuron-tracingstudyusingvirusvectors.J.Comp. Neurol.525,166–185

15. Ray,J.P.andPrice,J.L.(1993)Theorganizationofprojections fromthemediodorsalnucleusofthethalamustoorbitaland medialprefrontalcortexinmacaquemonkeys.J.Comp.Neurol. 337,1–31

16. Klein,J.C.etal.(2010)Topographyofconnectionsbetween humanprefrontalcortexandmediodorsalthalamusstudiedwith diffusiontractography.Neuroimage51,555–564

17. Usrey,W.M.andSherman,S.M.(2018)Corticofugalcircuits: communicationlinesfromthecortextotherestofthebrain.J. Comp.Neurol.PublishedonlineMarch10,2018.http://dx.doi. org/10.1002/cne.24423

18. Saalmann,Y.B.(2014)Intralaminarandmedialthalamicin flu-enceoncorticalsynchrony,informationtransmissionand cog-nition.Front.Syst.Neurosci.8,83

OutstandingQuestions

Howtopushformoreprecisionand accuracyinestimatesoftheMRIsignal (volumetric,diffusion,orbloodoxygen leveldependent)fromthemedial thal-amustoobtainmeasurementsthatare trulyspecifictoalterationsinthemedial thalamusandnotartifactualbecause ofnoise,lesions,oramixtureofsignal fromneighboringthalamicnuclei?This technical improvement would be importanttounderstandalterationsin pathologicalconditionslikeSCZand AUD.

Howcan the efficiency ofMD–PFC interactionsbequantifiedinhumans forthedevelopmentofneuroimaging targetsfortherapyinclinicalconditions characterizedbyMDdysfunction? Whatisthecognitiveadvantageofthe trans-thalamicrouteofcorticocortical communication?Arespecific triangu-larcircuitsinvolvingtheMDmediating specificcognitiveoperations?

HowdoesMDdysfunctionaffect tem-poral aspects of performance, as assessed,forexample,by implement-ingvariableresponsedeadlines, inter-ference atvariable times along the task, and variable response delays overseveralminutes?

IstheMDspecificallyrelatedto multi-tasking and interference management?

What is the cognitive complaint of patientswith MDlesionsinthelong run?

Whatarethebrain-widealterationsin gray andwhitematterfollowing MD lesions?

19. Saalmann,Y.B.etal.(2012)Thepulvinarregulatesinformation transmission between cortical areas based on attention demands.Science337,753–756

20. Guo,Z.V.etal.(2017)Maintenanceofpersistentactivityina frontalthalamocorticalloop.Nature545,181–186 21. Parnaudeau,S.etal.(2018)Themediodorsalthalamus:an

essentialpartneroftheprefrontalcortexforcognition.Biol. Psychiatry83,648–656

22. Acsady,L.(2017)Thethalamicparadox.Nat.Neurosci.20, 901–902

23. Wang,X.J.(2001)Synapticreverberationunderlyingmnemonic persistentactivity.TrendsNeurosci.24,455–463

24. Arcelli,P.etal.(1997)GABAergicneuronsinmammalian thala-mus:amarkerofthalamiccomplexity?BrainRes.Bull.42, 27–37

25. Garcia-Cabezas,M.A.etal.(2007)Distributionofthedopamine innervationinthemacaqueandhumanthalamus.Neuroimage 34,965–984

26. Garcia-Cabezas,M.A.etal.(2009)Dopamineinnervationinthe thalamus:monkeyversusrat.Cereb.Cortex19,424–434 27. Peters,S.K.etal.(2016)Cortico-striatal-thalamicloopcircuitsof

thesaliencenetwork:acentralpathwayinpsychiatricdisease andtreatment.Front.Syst.Neurosci.10,104

28. Zikopoulos,B.andBarbas,H.(2012)Pathwaysforemotions andattentionconvergeonthethalamicreticularnucleusin primates.J.Neurosci.32,5338–5350

29. Mitchell,A.S.(2015)Themediodorsalthalamusasahigher orderthalamicrelaynucleusimportantforlearningand deci-sion-making.Neurosci.Biobehav.Rev.54,76–88 30. Victor,M.A.etal.(1989)TheWernicke-Korsakoffsyndromeand

RelatedNeurologicDisorderduetoAlcoholismandMalnutrition, F.A.Davis

31. Victor,M.etal.(1971)TheWernicke-Korsakoffsyndrome.A clinicalandpathologicalstudyof245patients,82with post-mortemexaminations.Contemp.Neurol.Ser.7,1–206 32. Schmahmann,J.D.(2003)Vascularsyndromesofthethalamus.

Stroke34,2264–2278

33. JimenezCaballero,P.E.(2010)Bilateralparamedianthalamic arteryinfarcts:reportof10cases.J.StrokeCerebrovasc.Dis. 19,283–289

34. Pergola,G.etal.(2013)Quantitativeassessmentofchronic thalamicstroke.Am.J.Neuroradiol.34,E51–E55 35. Pergola,G.etal.(2013)Theinvolvementofthethalamusin

semanticretrieval:aclinicalgroupstudy.J.Cogn.Neurosci.25, 872–886

36. Danet,L.etal.(2015)Thalamicamnesiaafterinfarct:theroleof themammillothalamictractandmediodorsalnucleus. Neurol-ogy85,2107–2115

37. Pergola,G.etal.(2012)Recalldeficitsinstrokepatientswith thalamiclesionscovarywithdamagetotheparvocellular medi-odorsal nucleus of the thalamus. Neuropsychologia 50, 2477–2491

38. Zoppelt,D.etal.(2003)Involvementofthemediodorsalthalamic nucleusinmediatingrecollectionandfamiliarity. Neuropsycho-logia41,1160–1170

39. Parnaudeau,S.etal.(2013)Inhibitionofmediodorsalthalamus disruptsthalamofrontalconnectivityandcognition.Neuron77, 1151–1162

40. Watanabe,Y.andFunahashi,S.(2012)Thalamicmediodorsal nucleusandworkingmemory.Neurosci.Biobehav.Rev.36, 134–142

41. Watanabe,Y.andFunahashi,S.(2018)Changeofinformation representedbythalamicmediodorsalneuronsduringthedelay period.Neuroreport29,466–471

42. Newsome,R.N.etal.(2018)Dissociablecontributionsof tha-lamicnucleitorecognitionmemory:novelevidencefromacase ofmedialdorsalthalamicdamage.Learn.Mem.25,31–44

43. Kane,M.J.etal.(2007)Workingmemory,attentioncontrol,and theN-backtask:aquestion of constructvalidity. J.Exp. Psychol.Learn.Mem.Cogn.33,615–622

44. Carlesimo,G.A.etal.(2011)Prospectivememoryinthalamic amnesia.Neuropsychologia49,2199–2208

45. Cona,G.etal.(2018)Deficitsinprospectivememoryfollowing damagetothemedialsubdivisionofthemediodorsalthalamic nucleus.J.Neuropsychol.PublishedonlineMarch31,2018. http://dx.doi.org/10.1111/jnp.12154

46. Carlesimo,G.A.etal.(2011)Vascularthalamicamnesia:a reappraisal.Neuropsychologia49,777–789

47. Lovstad,M.etal.(2012)Executivefunctionsafterorbitalor lateral prefrontal lesions: neuropsychological profiles and self-reportedexecutivefunctionsineverydayliving.BrainInj. 26,1586–1598

48. VanderWerf,Y.D.etal.(2003)Contributionsofthalamicnuclei todeclarativememoryfunctioning.Cortex39,1047–1062 49. Ramon,M.etal.(2011)Thespeedofrecognitionofpersonally

familiarfaces.Perception40,437–449

50. Besson,G.etal.(2015)Fast,butnotslow,familiarityis pre-servedinpatientswithamnesticmildcognitiveimpairment. Cortex65,36–49

51. Tu,S.etal.(2014)Acceleratedforgettingofcontextualdetails duetofocalmedio-dorsalthalamiclesion.Front.Behav. Neuro-sci.8,320

52. Harding,A.etal.(2000)Degenerationofanteriorthalamicnuclei differentiatesalcoholicswithamnesia.Brain123,141–154 53. Pitel,A.L.etal.(2012)Macrostructuralabnormalitiesin

Korsak-off syndrome compared with uncomplicated alcoholism. Neurology78,1330–1333

54. Oscar-Berman,M.etal.(2014)Profilesofimpaired,spared,and recoveredneuropsychologicprocessesinalcoholism.Handb. Clin.Neurol.125,183–210

55. Lesh,T.A.etal.(2011)Cognitivecontroldeficitsin schizophre-nia:mechanismsandmeaning.Neuropsychopharmacology36, 316–338

56. Pergola,G.andSuchan,B.(2013)Associativelearningbeyond themedialtemporallobe:manyactorsonthememorystage. Front.Behav.Neurosci.7,162

57. Byne,W.etal.(2009)Thethalamusandschizophrenia:current statusofresearch.ActaNeuropathol.117,347–368 58. Dorph-Petersen,K.A. andLewis, D.A.(2017) Postmortem

structuralstudiesofthethalamusinschizophrenia.Schizophr. Res.180,28–35

59. vanErp,T.G.etal.(2016)Subcorticalbrainvolume abnormali-tiesin2028individualswithschizophreniaand2540healthy controlsviatheENIGMAconsortium.Mol.Psychiatry21,585 60. Ramsay,I.S.etal.(2018)Responsetotargetedcognitive train-ingcorrelateswithchangeinthalamicvolumeinarandomized trialfor earlyschizophrenia.Neuropsychopharmacology43, 590–597

61. Pergola,G.etal.(2015)Theroleofthethalamusin schizophre-niafromaneuroimagingperspective.Neurosci.Biobehav.Rev. 54,57–75

62. Pergola,G.etal.(2017)Greymattervolumepatternsinthalamic nucleiareassociatedwithfamilialriskforschizophrenia. Schiz-ophr.Res.180,13–20

63. Cobia,D.J.etal.(2017)Progressivedeteriorationofthalamic nuclei relatestocortical networkdecline in schizophrenia. Schizophr.Res.180,21–27

64. Weinberger,D.R.andRadulescu,E.(2016)Findingtheelusive psychiatriclesionwith21st-centuryneuroanatomy:anoteof caution.Am.J.Psychiatry173,27–33

66. Welsh,R.C.etal.(2010)Low-frequencyBOLDfluctuations demonstratealteredthalamocorticalconnectivityin schizophre-nia.Schizophr.Bull.36,713–722

67. Woodward,N.D.etal.(2012)Thalamocorticaldysconnectivityin schizophrenia.Am.J.Psychiatry169,1092–1099 68. Anticevic,A.etal.(2013)Characterizingthalamo-cortical

dis-turbancesinschizophreniaandbipolarillness.Cereb.Cortex 24,3116–3130

69. Antonucci,L.A.et al.(2016)Association offamilialriskfor schizophreniawiththalamicandmedialprefrontalfunctional connectivityduringattentionalcontrol.Schizophr.Res.173, 23–29

70. Woodward,N.D.andHeckers,S.(2016)Mapping thalamocort-icalfunctionalconnectivityinchronicandearlystagesof psy-choticdisorders.Biol.Psychiatry79,1016–1025

71. Anticevic,A.etal.(2015)Associationofthalamic dysconnec-tivityandconversiontopsychosisinyouthandyoungadultsat elevatedclinicalrisk.JAMAPsychiatry72,882–891 72. Ramsay,I.S.andMacDonald,A.W.,3rd (2018)Theupsand

downsofthalamocorticalconnectivityinschizophrenia.Biol. Psychiatry83,473–474

73. Pergola,G.etal.(2013)Theroleofthethalamicnucleiin recognitionmemoryaccompaniedbyrecallduringencoding andretrieval:anfMRIstudy.Neuroimage74,195–208 74. Kafkas,A.andMontaldi,D.(2014)Twoseparate,but

interact-ing,neuralsystemsforfamiliarityandnoveltydetection:a dual-routemechanism.Hippocampus24,516–527

75. Kafkas,A.andMontaldi,D.(2012)Familiarityandrecollection producedistincteyemovement,pupilandmedialtemporallobe responseswhenmemorystrengthismatched. Neuropsycho-logia50,3080–3093

76. Montaldi,D.etal.(2006)Theneuralsystemthatmediates familiaritymemory.Hippocampus16,504–520

77. Bartra,O.etal.(2013)Thevaluationsystem:a coordinate-basedmeta-analysisofBOLDfMRIexperiments examining neuralcorrelatesofsubjectivevalue.Neuroimage76,412–427 78. Balleine,B.W.etal.(2015)Thalamocorticalintegrationof instru-mentallearningandperformanceandtheirdisintegrationin addiction.BrainRes.1628,104–116

79. deBourbon-Teles,J.etal.(2014)Thalamiccontrolofhuman attention driven by memory andlearning. Curr. Biol. 24, 993–999

80. Metzger,C.D.etal.(2010)HighfieldFMRIreveals thalamocort-icalintegrationofsegregatedcognitiveandemotional process-ing in mediodorsal andintralaminar thalamic nuclei.Front. Neuroanat.4,138

81. Rosen,M.L.etal.(2017)Corticalandsubcorticalcontributions tolong-termmemory-guidedvisuospatialattention.Cereb. Cor-tex28,2935–2947

82. Zotev,V.etal.(2018)Real-timefMRIneurofeedbackofthe mediodorsal and anterior thalamus enhances correlation between thalamic BOLD activity andalpha EEG rhythm. Hum.BrainMapp.39,1024–1042

83. Nakajima,M.andHalassa,M.M.(2017)Thalamiccontrolof functional cortical connectivity. Curr. Opin.Neurobiol. 44, 127–131

84. Staudigl,T.etal.(2012)Memorysignalsfromthethalamus:early thalamocorticalphasesynchronizationentrainsgamma oscilla-tionsduringlong-termmemoryretrieval.Neuropsychologia50, 3519–3527

85. Sweeney-Reed,C.M.etal.(2016)Pre-stimulusthalamictheta powerpredictshumanmemoryformation.Neuroimage138, 100–108

86. Perakyla,J.etal.(2017)Causalevidencefromhumansforthe roleofmediodorsalnucleusofthethalamusinworkingmemory. J.Cogn.Neurosci.29,2090–2102

87. Sherman,S.M.andGuillery,R.W.(2002)Theroleofthe thala-musintheflowofinformationtothecortex.Philos.Trans.R. Soc.Lond.BBiol.Sci.357,1695–1708

88. Sherman,S.M.andGuillery,R.W.(2006)Exploringthe Thala-musandItsRoleinCorticalFunction,MITPress

89. Sherman,S.M.(2007)Thethalamusismorethanjustarelay. Curr.Opin.Neurobiol.17,417–422

90. Sherman,S.M.andGuillery,R.W.(2011)Distinctfunctionsfor directandtransthalamiccorticocorticalconnections.J. Neuro-physiol.106,1068–1077

91. LaBerge,D.(1997)Attention,awareness,andthetriangular circuit.Conscious.Cogn.6,149–181

92. Schneegans,S.andBays,P.M.(2018)Driftinneuralpopulation activitycausesworkingmemorytodeteriorateovertime.J. Neurosci.38,4859–4869

93. Baddeley,A.(2000)Theepisodicbuffer:anewcomponentof workingmemory?TrendsCogn.Sci.4,417–423

94. Jones,E.G.(2007)TheThalamus,CambridgeUniversityPress 95. Huguenard,J.R.andMcCormick,D.A.(2007)Thalamic syn-chronyanddynamicregulationofglobalforebrainoscillations. TrendsNeurosci.30,350–356

96. Rikhye,R.V.etal.(2018)Towardanintegrativetheoryof tha-lamicfunction.Annu.Rev.Neurosci.41,163–183 97. Zikopoulos,B.andBarbas,H.(2006)Prefrontalprojectionsto

thethalamicreticularnucleusformauniquecircuitforattentional mechanisms.J.Neurosci.26,7348–7361

98. Lezak,M.D.(1982)Theproblemofassessingexecutive func-tions.Int.J.Psychol.17,281–297

99. Danet,L.etal.(2017)Medialthalamicstrokeanditsimpacton familiarityandrecollection.eLife6,e28141

100.Cipolotti,L.etal.(2008)Theroleofthethalamusinamnesia:a tractography, high-resolution MRI and neuropsychological study.Neuropsychologia46,2745–2758

101.Soei,E.etal.(2008)Involvementofthehumanthalamusin relationalandnon-relationalmemory.Eur.J.Neurosci.28, 2533–2541

102.Aggleton,J.P.etal.(2011)Unravelingthecontributionsofthe diencephalontorecognitionmemory:areview.Learn.Mem.18, 384–400

103.Edelstyn,N.M.etal.(2016)Adeficitinfamiliarity-driven recog-nitioninaright-sidedmediodorsalthalamiclesionpatient. Neu-ropsychology30,213–224

104.Morel,A.(2007)StereotacticAtlasoftheHumanThalamusand BasalGanglia,CRCPress

105.Kanowski,M.etal.(2014)Directvisualizationofanatomic sub-fieldswithinthesuperioraspectofthehumanlateralthalamusby MRIat7T.Am.J.Neuroradiol.35,1721–1727

106.Deistung,A.etal.(2013)Towardinvivohistology:acomparison ofquantitativesusceptibilitymapping(QSM)withmagnitude-, phase-,andR2*-imagingatultra-highmagneticfieldstrength. Neuroimage65,299–314

107.Tourdias,T.etal.(2014)Visualizationofintra-thalamicnuclei withoptimizedwhite-matter-nulledMPRAGEat7T.Neuroimage 84,534–545

108.Krauth,A.etal.(2010)Ameanthree-dimensionalatlasofthe humanthalamus:generationfrommultiplehistologicaldata. Neuroimage49,2053–2062

109.Pergola,G.(2016)Thalamicamnesiaafterinfarct:theroleofthe mammillothalamictractandmediodorsalnucleus.Neurology 86,1928

110.Behrens,T.E.J.etal.(2007)Probabilisticdiffusiontractography withmultiplefibreorientations:whatcanwegain?Neuroimage 34,144–155

111.Behrens,T.E.etal.(2003)Non-invasivemappingofconnections betweenhumanthalamusandcortexusingdiffusionimaging. Nat.Neurosci.6,750–757

113.O’Muircheartaigh,J.etal.(2015)Whitematterconnectivityof thethalamusdelineatesthefunctionalarchitectureofcompeting thalamocorticalsystems.Cereb.Cortex25,4477–4489 114.Jbabdi,S.etal.(2009)Multiple-subjectsconnectivity-based

parcellationusinghierarchicalDirichletprocessmixturemodels. Neuroimage44,373–384

115.Hale,J.R.etal.(2015)Comparisonoffunctionalthalamic seg-mentationfromseed-basedanalysisandICA.Neuroimage114, 448–465

116.Zhang,D.etal.(2010)Noninvasivefunctionalandstructural connectivitymappingofthehumanthalamocorticalsystem. Cereb.Cortex20,1187–1194

117.Ji,B.etal.(2016)Dynamicthalamusparcellationfrom resting-statefMRIdata.Hum.BrainMapp.37,954–967

118.vanOort,E.S.B.etal.(2018)Functionalparcellationusingtime coursesofinstantaneousconnectivity.Neuroimage170,31–40 119.Kumar,V.J.etal.(2017)Functionalanatomyofthehuman

thalamusatrest.Neuroimage147,678–691

120.Johansen-Berg,H.etal.(2005)Functional-anatomicalvalidation andindividualvariationofdiffusiontractography-based seg-mentationofthehumanthalamus.Cereb.Cortex15,31–39 121.Shallice,T.andBurgess,P.W.(1991)Deficitsinstrategy

appli-cationfollowingfrontallobedamageinman.Brain114,727–741 122.Petrides,M.andMilner,B.(1982)Deficitsonsubject-ordered tasksafterfrontal-andtemporal-lobelesionsinman. Neuro-psychologia20,249–262

123.Quinette,P.etal.(2006)Therelationshipbetweenworking memoryandepisodicmemorydisordersin transientglobal amnesia.Neuropsychologia44,2508–2519

124.Chakraborty,S.etal.(2018)Macaqueparvocellular mediodor-salthalamus:dissociablecontributionstolearningandadaptive decision-making. Eur. J. Neurosci. Published online July 18,2018.http://dx.doi.org/10.1111/ejn.14078

125.Kopelman,M.D.etal.(2009)TheKorsakoffsyndrome:clinical aspects, psychology and treatment. Alcohol Alcohol. 44, 148–154

126.Pitel,A.L.etal.(2015)Thalamicabnormalitiesareacardinal featureofalcohol-relatedbraindysfunction.Neurosci. Biobe-hav.Rev.54,38–45

127.Pfefferbaum,A.etal.(2014)Whitemattermicrostructural recov-erywithabstinenceanddeclinewithrelapseinalcohol depen-denceinteractswithnormalageing:acontrolledlongitudinalDTI study.LancetPsychiatry1,202–212

128.Arts,N.J.etal.(2017)Korsakoff’ssyndrome:acriticalreview. Neuropsychiatr.Dis.Treat.13,2875–2890

129.Segobin,S.H.etal.(2014)Relationshipbetweenbrain volumet-ricchangesandinterimdrinkingatsixmonthsin alcohol-depen-dentpatients.Alcohol.Clin.Exp.Res.38,739–748 130.Savage,L.M.etal.(2012)Translationalrodentmodelsof

Kor-sakoffsyndromerevealthecriticalneuroanatomicalsubstrates ofmemorydysfunctionandrecovery.Neuropsychol.Rev.22, 195–209

131.Vedder,L.C.etal.(2015)Interactionsbetweenchronicethanol consumptionandthiaminedeficiencyonneuralplasticity,spatial memory,andcognitiveflexibility.Alcohol.Clin.Exp.Res.39, 2143–2153

132.Qin,L.andCrews,F.T.(2014)Focalthalamicdegenerationfrom ethanolandthiaminedeficiencyisassociatedwithneuroimmune geneinduction,microglialactivation,andlackof monocarbox-ylicacidtransporters.Alcohol.Clin.Exp.Res.38,657–671 133.Ritz,L.etal.(2016)Clinicalandbiologicalriskfactorsfor

neuro-psychologicalimpairmentinalcoholusedisorder.PLoSOne11, e0159616

134.Pitel,A.L.etal.(2011)SignsofpreclinicalWernicke’s encepha-lopathyandthiaminelevelsaspredictorsofneuropsychological deficitsinalcoholismwithoutKorsakoff’ssyndrome. Neuropsy-chopharmacology36,580–588