2005b The primate amygdala neuronal representations of the viscosity, fat texture, grittine

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

Neuroscience132(2005)33–48

THEPRIMATEAMYGDALA:NEURONALREPRESENTATIONSOFTHEVISCOSITY,FATTEXTURE,TEMPERATURE,GRITTINESSANDTASTEOFFOODS

M.KADOHISA,J.V.VERHAGENANDE.T.ROLLS*

UniversityofOxford,DepartmentofExperimentalPsychology,SouthParksRoad,OxfordOX13UD,UK

Abstract—Theprimateamygdalaisimplicatedinthecontrolofbehavioralresponsestofoodsandinstimulus-reinforcementlearning,butonlyitstasterepresentationoforalstimulihasbeeninvestigatedpreviously.Of1416macaqueamygdalaneu-ronsrecorded,44(3.1%)respondedtooralstimuli.Ofthe44orallyresponsiveneurons,17(39%)representtheviscosityoforalstimuli,testedusingcarboxymethyl-celluloseintherange1–10,000cP.Twoneurons(5%)respondedtofatinthemouthbyencodingitstexture(shownbytheresponsesoftheseneu-ronstoarangeoffats,andalsotonon-fatoilssuchassiliconeoil((Si(CH3)2O)n)andmineraloil(purehydrocarbon),butnoorsmallresponsestothecelluloseviscosityseriesortothefattyacidslinoleicacidandlauricacid).Ofthe44neurons,three(7%)respondedtogrittytexture(producedbymicrospheressus-pendedincellulose).Eighteenneurons(41%)respondedtothetemperatureofliquidinthemouth.Someamygdalaneuronsrespondedtocapsaicin,andsometofattyacids(butnottofatsinthemouth).Someamygdalaneuronsrespondtotaste,tex-tureandtemperatureunimodally,butotherscombinetheseinputs.Theseresultsprovidefundamentalevidenceabouttheinformationchannelsusedtorepresentthetextureand avoroffoodinapartofthebrainimportantinappetitiveresponsestofoodandinlearningassociationstoreinforcingoralstimuli,andarerelevanttounderstandingthephysiologicalandpathophys-iologicalprocessesrelatedtofoodintake,foodselection,andtheeffectsofvarietyoffoodtextureincombinationwithtasteandotherinputsonfoodintake.©2005PublishedbyElsevierLtdonbehalfofIBRO.

Keywords:appetitivelearning,Pavlovianconditioning,obesity,foodintake,reward.

Theamygdalaisimplicatedinemotionandmotivationbylesion,singleneuronrecording,andneuroimaginginvestiga-tions(Sangheraetal.,1979;Nishijoetal.,1988a;Davis,1994;Francisetal.,1999;Rolls,1999;Schoenbaumetal.,1999;LeDoux,2000;Rolls,2000).Partofthefunctionoftheamygdalainmotivationandemotionappearstobeinasso-ciatingpreviouslyneutral(e.g.auditoryorvisual)stimulitoprimary(unlearned)reinforcerssuchastasteandsomato-sensoryincludingpainfulstimuli(Davis,1994;Rolls,1999,2000;LeDoux,2000),andinprimates,forexample,rewarddevaluationlearningdependsontheamygdala(Murrayetal.,

*Correspondingauthor.Tel: 44-1865-271348;fax: 44-1865-310447.E-mailaddress:edmund.rolls@psy.ox.ac.uk(E.T.Rolls).

Abbreviations:BJ,blackcurrantjuice;CMC,carboxymethylcellulose;CO,coconutoil;GR,grittytexturestimulus;LaA,lauricacid;LiA,linoleicacid;MO,mineraloil;SaO,saf oweroil;SC,singlecream;SiO,siliconeoil;VO,vegetableoil.

0306-4522/05$30.00 0.00©2005PublishedbyElsevierLtdonbehalfofIBRO.doi:10.1016/j.neuroscience.2004.12.005

1996).Thisassociativefunction(RollsandTreves,1998)impliesarepresentationofprimaryreinforcerssuchasthetasteoffood,andindeedtasteneuronshavebeendescribedintheamygdala(Sangheraetal.,1979;Nishijoetal.,1988a,b;Scottetal.,1993;YanandScott,1996)andtasteresponsivenessinthehumanamygdala(O’Dohertyetal.,2001).Almostnothingisknown,however,aboutwhetheraspectsoffoodotherthantasteandsmellarerepresentedintheamygdala.Thetextureoffoodisimportantinitspalatabil-ityandacceptability(Bourne,2002,considere.g.dampce-realorpotatochips),andtemperaturemayalsobeimportant(Zellneretal.,1988).Wedescribehereforthe rsttimetheresponsesofprimateamygdalaneuronstooraltextureandtemperaturestimuli,andshowthattheyarecombinedinsomeneuronswithresponsivenesstothetasteoffood.Theinvestigationwasperformedinmacaquesinordertomakeitasrelevanttounderstandingtheoperationofthissysteminhumansaspossible,andinthecontextthatthetastesystemisdifferentlyconnectedinrodentsandprimates,withtastepathwaysinprimatestotheamygdalaonlyaftercorticalprocessing(Norgren,1984;Rollsetal.,2003).Understand-ingthefactorsthatdeterminethepalatabilityoffoodiscur-rentlyofgreatinterest,giventheroleofpalatabilityinthecontroloffoodintake,andtherapidlyincreasingincidenceofobesitywhichisaccompaniedbyserioushealthrisks(Berthoud,2003;SteinbergerandDaniels,2003).Thefactorsinvestigatedincludedthetextureoffoodasre ectedbyvis-cosity(testedparametricallywithaviscosityseriesmadewithcellulose);oralfat;oralfattyacidswhichmightsignalthepresenceoffatinthemouthifsalivarylipasewaspresent(Gilbertson,1998);oraltextureasmanipulatedbyinertmi-crospheres;oraltemperature;taste;andtheeffectsofcap-saicin,anoralirritantpresentinanumberoffoods.

EXPERIMENTALPROCEDURES

Subjects

Therecordingsweremadeinthreehemispheresoftworhesusmacaques(Macacamulatta;onefemaleweighing2.6–3.3kgandonemaleweighing6.1–6.7kg).Toensurethatthemacaqueswerewillingtoingestthetestfoodsand uidsduringtherecordingsessions,theywereonmildfood(150gofnutritionallybalancedmashplusfruits,boiledchickeneggs,nuts,seedsandpopcorn)and uid(1h/dayadlibitumwater)deprivation,inthatbothwereprovidedafterthedailyrecordingsession.Themonkeysshowedsteadyincreasesinbodyweight.Allprocedures,includingprepar-ativeandsubsequentones,werecarriedoutinaccordancewiththeNationalInstitutesforHealthGuidefortheCareandUseofLaboratoryAnimals,andwerelicensedundertheUKAnimals(Scienti cProcedures)Act,1986,andweredesignedtominimize

33

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

34M.Kadohisaetal./Neuroscience132(2005)33–48

thenumberofanimalsusedandmaximizetheirwelfarebyadopt-ing,forexample,grouphousing,andenvironmentalenrichments.

Recordings

Recordingsweremadewithepoxylite-coatedhighimpedance(2–10M at1kHz;FredericHaer&Co.,St.Bowdoinham,ME,USA)tungstenmicroelectrodesfromsingleneuronsintheamyg-dala,whichincludedareasinwhichtasteresponseshaveprevi-ouslybeendescribed(Sangheraetal.,1979;Nishijoetal.,1988a,b;Scottetal.,1993),usingneurophysiologicalmethodsasdescribedpreviously(Scottetal.,1986a,b;Rollsetal.,1990,1999,2003;Verhagenetal.,2003).Thesignal-to-noiseratiowastypically3:1orhigherasillustratedinFig.4.ThedatawerecollectedusingaDatawaveDiscoveryInc.(Tucson,AZ,USA)systemwhichdigitizedthesignal(12bit,16kHz)for8safterstimulusonset.Thespikesweresortedoff-lineusingtheclustercuttingmethodprovidedwiththeDatawavesystem,andthisprocedurewasstraightforwardasthedatawerecollectedwithsingleneuronmicroelectrodeswhichtypicallyrecordedfromonlyoneneuronatatime.Topreventvisualassociativeinputfromevokingneuralactivity,wepreventedthemonkeysfromseeingthestimuliandexperimenterbyaview-obstructingscreen.

Localizationofrecordings

X-radiographywasusedtodeterminethepositionofthemicro-electrodeaftereachrecordingtrackrelativetopermanentrefer-enceelectrodesandtotheanteriorsphenoidalprocess,followedbymicrolesionsonselectedtracksandreconstructionfromhisto-logicalsectionsofthebrainusingthemethodsdescribedbyFeigenbaumandRolls(1991).

Stimuli

Theneuronsoftheamygdalaweretestedfortheirresponsivenesstothesetoftaste,viscosity,gritty,oilystimuli,andcapsaicin,atroomtemperature(23°C),andalsothesetoftemperaturestimuliasshowninTable1.DetailsoftherationaleforthechoiceofthestimuliaregivenbyRollsetal.(2003)andVerhagenetal.(2003).Distilledwaterat23°Cwasonememberofthetemperatureseries(T23),andwithitsviscosityof1cPwasalsoonemember(V1)oftheviscosityseries.Foranadditionalcomparison,theneuronalresponsesweretestedto20%blackcurrantjuice(BJ;RibenaSmithKlineBeecham),becausewithitscomplextasteandolfac-torycomponentsandhighpalatabilityitisaneffectivestimuluswhensearchingforandanalyzingtheresponsesofcorticalneu-rons(Rollsetal.,1990).Theviscosityserieswasmadewithcarboxymethylcellulose(CMC;Sigma;highviscosity,Mw700,000,dialysed,codeC5013),avirtuallyodor-andtastelessthickeningagentusedwidelyinthefoodindustry(seeRollsetal.,2003).Thegrittystimulusconsistedofhard(Mohsscale5)hollowmicrospheres(FillitegradePG,with87%havingadiameterwiththerange100–300 m;TrelleborgFillite,Runcorn,UK)madeupinmethylcellulosetohaveameasuredviscosityof1000cP(100gofFillitePGwasaddedto4.7gofCMCin500mlofwater).Totestforandanalyzetheeffectsoforalfatonneuronalactivity,asetofoilsandfat-relatedstimuliwasincluded.Thetriglyceride-basedoilsconsistedofvegetableoil(VO;viscosity55cPat23°C),saf oweroil(SaO),andcoconutoil(CO).Singlecream(SC;18%fat,viscosity:12cP;Coopbrand,pasteurized)wasusedasanexemplarofanaturalhighfatcontentfoodofthetypeforwhichwewishedtoexaminetheneuralrepresentationandsensingmech-anisms.Fouroftheorallyresponsiveneuronsweretestedwithmineraloil(MO),ahydrocarbonmixturewithaviscosityof25cP.Alltheneuronswithfat-relatedresponsesdescribedinthisandourearlierstudy(Rollsetal.,1999)respondedwelltoSC.AsGilbertson(1998)hadreporteddifferentialeffectsinisolatedtastecellstolinoleicandlauricacid(LaA)invitro,suggestingthatthe

gustatorymodalitymightbeinvolvedinorallysensingfat,weincluded(Verhagenetal.,2003)inthestimulussetfreelinoleic(LiA;100 M)andLaA(100 M)sodiumsalts(Sigma),aswellasoilsrichinconjugatedLiA(SaO,68–83%,50cP;Aldrich),andLaA(CO,45–50%,40cP;Sigma;Weiss,1983;Willsetal.,1998).AllfattyoilswerekeptinthedarkunderN2at4°Ctoavoidoxidation.

Toinvestigatewhethertheneuronsresponsivetofat-basedoilswereinsomewayrespondingtothesomatosensorysensationselicitedbythefat,stimuliwithasimilarmouthfeelbutnon-fatchem-icalcompositionwereused.Thesestimuliincludedparaf n/MO(purehydrocarbon,viscosity25cPat23°C,Sigma),andsiliconeoil(SiO;Si(CH3)2O)n,10,100or280,and1000cP(Brook eldviscometercalibration uidexcept280cP;Aldrich).

Thetemperatureserieswasprovidedbywaterat10°C(chosenasthecoldstimulus;commercialcolddrinksareservedat6°C),at42°C(warm/hotbutnotnoxious),37°C(bodytemper-ature),and23°C(roomtemperature).

Thecapsaicinwasmadeupasa10 Msolution(containing0.3%ethanol).Thisisapproximately15timesthehumanrecog-nitionthresholdof0.66 M(Szolcsanyi,1990).

Themonkeys’preferenceforthestimuliwasmeasuredob-jectivelybyanacceptabilityrating,where 2indicatesthatthemacaquereachedforthestimulusandplaceditinthemouth, 1indicatesthatthemacaqueactivelyopenedthemouthtoreceiveandswallowthestimulus,0indicatesneutralityinwhichthestim-uluswasacceptedonlypassivelybymouthopening, 1indicatesthatthemacaqueclosedthemouthtotrytorejectthestimulus;and 2indicatesthatthemacaqueusedthehandtopushawaythestimulusfromthemouth(Rollsetal.,1977,1989).Thisratingscalehasbeenextensivelyvalidatedbycomparisonwithneuronalactivityinthelateralhypothalamusandorbitofrontalcortexinstudiesofsensory-speci csatiety,inthesensethattherewasacloserelationfoundbetweentheacceptabilityratingandtheneu-ronalresponseintheseregions,asshowninpreviouslypublisheddata(Rollsetal.,1986,1989;CritchleyandRolls,1996).Theratingswererepeatedatotaloffour(orforonemacaque5)timesbytwoindependentinvestigators,withahighcorrelationbetweentheratingsgivenbythetwoinvestigators(r 0.85,n 25,P 10 7).

Stimulusdelivery

Thegeneralmethodforstimulusdeliveryandaccuratestimulusonsetmarking(Rollsetal.,1990)wasmodi edbyintroducingrepeaterpipettes(Verhagenetal.,2003).Forchronicrecordinginmonkeys,amanualmethodforstimulusdeliveryisusedbecauseitallowsforrepeatedstimulationofalargereceptivesurfacedespitedifferentmouthandtonguepositionsadoptedbythemon-keys(Scottetal.,1986a,b).Thestimulusapplicationvolumewas200 10 l,becausethisissuf cienttoproducelargegustatoryneuronalresponseswhichareconsistentfromtrialtotrial,andyetwhichdonotresultinlargevolumesof uidbeingingestedwhichmight,byproducingsatiety,in uencetheneuronalresponses(Rollsetal.,1989,1990).

Themonkey’smouthwasrinsedwith200 lT23/V1(water)duringtheinter-trialinterval(whichlastedatleast30s,oruntilneuronalactivityreturnedtobaselinelevels)betweentastestim-uli.Duetothetenaciousnatureoftheoralcoatingresultingfromthedeliveryofcreamorofoil,andalsoforgrittyandcapsaicin,four200 l-rinseswithT23/V1weregiven,whileallowingthesubjectstoswallowaftereachrinse.ForV1000andV10,000,weusedtwosuchrinses.AllthestimulishowninTable1weredeliveredinpermutedsequences,withthecomputerspecifyingthenextstimulustobeusedbytheexperimenter.Thespontane-ous ringrateoftheneuronwasmeasuredfromtrialsinwhichnostimulusdeliveryoccurred.

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–48

Table1.StimuliStimulus

Abbrev

Concentration

MW

Temp(°C)

180

2323

1875836387

2323232310233742232323232323232323232323232323

Visco(cP)

1111111111101001000100001000

10100Or2801000

55405012111

Chemicalgroup

Acceptabilitybkmean S.E.M.1.8 0.1*1.8 0.1* 0.7 0.7 1.6 0.1*0.40 0.480.10 0.330.60 0.370.06 0.13*1.50 0.201.46 0.230.05 0.52 0.76 0.28* 0.83 0.33* 1.56 0.26* 1.34 0.18*

—0.032 0.63

—

0.93 0.17 1.00 0.31 1.14 0.11*1.02 0.65* 0.13 0.680.00 0.17 1.10 0.11

35

Acceptabilitybomean S.E.M.1.3 0.31.0 0.0 0.6 0.3 0.8 0.1 0.13 0.830.63 0.431.25 0.291.00 0.001.00 0.001.25 0.291.00 0.001.00 0.001.00 0.001.00 0.001.25 0.29 1.13 0.34 1.13 0.341.00 0.00 1.38 0.42 0.13 0.830.63 0.43 1.69 0.36 0.50 1.000.25 0.870.25 0.87

Glucose

BlackcurrantjuiceMonosodiumglutamateNaClHCl

QuinineHClWaterWaterWaterWaterCMCCMCCMCCMCGrittySiliconeoilSiliconeoilSiliconeoilVegetableoilCoconutoilSaf oweroilSinglecreamLauricacidLinoleicacidCapsaicin

GBJMNHQT10T23/V1T37T42V10V100V1000V10000GrSiO10SiO100orSiO280SiO1000VOCOSaOSCLaALiACap

1M20%0.1M0.1M0.01M0.001M

MonosaccharidealdohexoseMixtureAminoacidsaltInorganicsaltInorganicacidAlkaloid

0.2g 1lV14.0g 1lV111.0g 1lV124.0g 1lV1100gFillite

9.4gCMC 1lV1100%100%100%100%100%100%100%100 M100 M10 M

700,000700,000700,000700,000700,000

222280377

PolysaccharidePolysaccharidePolysaccharidePolysaccharideSiO2

polysaccharideSilicon-oxygenpolymerSilicon-oxygenpolymerSilicon-oxygenpolymerFatFatFat

Emulsionffaffa

Vanillylamide

*P 0.05acceptabilityratingbkvsbo;n 4–5.NA,notapplicable.

Dataanalysis

AfterclustercuttingofthespikeswithDatawavesoftware,thenum-bersofspikesofthesingleneuronin80timebinseach100mslongstartingattheonsetofthestimuluswereobtainedusingSPSS.Statisticalanalysiswasperformedonthenumbersofspikesinthe rst1speriodafterstimulusonset,whichwassuf cientlylongtoinclude ringtoevenviscousliquids,andsuf cientlyshortsothatlowviscositytastestimuliwerestillactivatingtheneurons,asshowninFig.2ofRollsetal.(2003)andinFig.4.AnANOVAwasperformed(withSPSS)todeterminewhethertheneuronhadsigni cantlydif-ferentresponsestothesetofstimuli.IfthemainANOVAwassigni cant,fourfurtherANOVAswereperformedtotestfordiffer-encesinneuronalresponsesbetweenthesetoftastestimuli(G,N,H,Q,MandT23/V1),betweenthemembersoftheviscosityseriesV1–V10,000,thesetoffatstimuli(MO,SiO10,100or280,1000,VO,CO,SaO),andthesetoftemperaturestimuli(T10–T42).Systat10wasusedforthegenerationofPearsonproduct-momentcorrelationcoef cientscalculatedbetweenthestimuliusingtheresponsesofalltheneuronsanalyzed,andgraphicalpresentationofstimulussimi-larityusingmultidimensionalscaling(lossfunction:Kruskal;regres-sion:mono).

Atastecellwasde nedbyasigni canteffectintheANOVAperformedacrossthestimulussubset(V1,G,N,M,H,Q)onthenumberofspikesduringthe rstsecondafterstim-ulusonset.Similarly,theviscositycellcriterionwasbasedonasigni canteffectintheANOVAbetweenthesetofstimuliV1–V10,000.Fatcellswerede nedbyasigni cantlylargeraverage ringratetotheoils(viscosity25–100cP)thantothe

averageratestoV10andV100;andbyinadditionasigni cantlargeraverage ringratetotheoilsthanthespontaneous ringrate.Thecriterionforbeingsensitivetotemperaturewasbasedonasigni canteffectintheANOVAbetweenthesetofstimuliT10–T42.Thecritical levelwassetatP 0.05.Further,thetestsforcapsaicin,LaA,andLiAsensitivitywereatwo-tailedt-testcomparingtheresponsesoftheneurontocapsaicin,LaA,andLiAandtowater.Thetestforgrittytexturesensitivitywasatwo-tailedt-testcomparingtheresponsesoftheneurontothegrittytexturestimulus(Gr;whichhasaviscosityof1000cP)andtothe1000cPstimulusfromtheviscosityseriesmadewithCMC.

AFisher(1932)probabilitycombination(orgeneralizedsig-ni canceorexactprobability)testwasperformedtocheckthatthestatisticallysigni cantresultsintheorallyresponsivecellscouldnotre ectjustchancestatisticalresults.(Bychance,ifforexampleonestatisticaltestwasperformedon100cells,then veofthetestsmightbeexpectedtobesigni cantatP 0.05.)TheFishercombinationtestcalculatestheexactprobabilityofobtainingasetofsigni cancevaluesbychanceinindependenttests.Thepro-cedurecalculates 2Αlnpi,whichhasa 2distributionwith2ndegreesoffreedom,andthesumisoverthenprobabilityvaluespiobtainedinseparatetests.ThismeasureiswellestablishedandasymptoticallyBahaduroptimal(LittellandFolks,1971;Zaykinetal.,2002).

Toquantifythetuningoftheneuronstothestimuli,thebreadthoftuningmetricHofSmithandTravers(1979)wascalculatedas

H k

ipilogpi

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

36M.Kadohisaetal./Neuroscience132(2005)33–48

Fig.1.(A)Responsesofanamygdalaneuron(bo205)withdifferentialresponsestoviscosity,temperatureandtaste.Themean( S.E.M.) ringrateresponsestoeachstimuluscalculatedina1speriodoverfourtosixtrialsareshownhereandelsewhereunlessotherwiseindicated.Thespontaneous(Spon) ringrateisshown.Thetastestimuliwere1Mglucose(G),0.1MNaCl(N),0.1MMSG(M),0.01MHCl(H)and0.001MQuinine–HCl(Q);thetemperaturestimuliwereT10,T23,T37andT42wherethenumberindicatesthetemperaturein°C;theviscositystimuliwereV1,V10,V100,V1000andV10,000wherethenumeralindicatestheviscosityincPat23°C);fattexturestimuliwereSiO10,SiO100,SiO1000(SiOwiththeviscosityindicated),VO,COandSaO.BJisfruitjuice;Capis10 Mcapsaicin;LaAis0.1mMLaA;LiAis0.1mMLiA;seeTable1.(B)Theresponsesofanamygdalaneuron(bo219c2)respondingtotemperature.ConventionsasinFig.1A.

wherek scalingconstant(setsothatH 1.0whentheneuronrespondsequallywelltoallstimuliinthesetofsizen),pi theresponsetostimulusiexpressedasaproportionofthetotalre-sponsetoallthenstimuliintheset.Thecoef cientrangesfrom0.0,representingtotalspeci citytooneofthestimuli,to1.0,whichindicatesanequalresponsetoallofthestimuli.Thesparsenessoftherepresentationawasalsomeasurequantitativelybyextendingthebinarynotionoftheproportionofneuronsthatare ring,as

a (

Screeningcells

Whilesearchingforneurons,wecontinuouslyappliedsamplesfromourstimulusset:G,N,Q,BJ,SC,VO,SO,V100,V1/T23,T10,T42.Wealsotestedforvisualresponsiveness(tothesightoffood,asaline-associatedsquareplaque,theapproachofatastestimulustowardthemouth,objects,faces,headmovement,andlip-smacking)andauditoryresponsiveness(a500Hztone,coo-calls,gruntsandvocalization)asstimuliofthesetypesdoactivatesomeamygdalaneurons(Sangheraetal.,1979).Whenneuronswereinsensitivetothesestimuli,weclassi edthemasnon-responsive.Onlycellsrespondingconsistentlytoatleastonestimulusofthearraywererecorded,allstimulibeingappliedfourtosixtimesinpermutedsequences.

i 1,N

ri N)2

i 1,N(ri2 N)

whereriisthe ringrateoftheithneuroninthesetofNneurons(RollsandTovee,1995;RollsandTreves,1998;RollsandDeco,2002).Thesparsenessiswithintherange0–1,andassumesthevalue0.5forafullydistributedrepresentationwithbinaryencod-ing;and1/Nforalocalorgrandmothercellrepresentationwithbinaryencoding.Thesemeasuresofthe nenessofthetuningofneuronsareimportantinunderstandingtheneuronalencodingofinformation(RollsandTreves,1998;RollsandDeco,2002).

RESULTS

Thedatadescribedinthispaperwereobtainedinthreehemispheresoftwomonkeys.Outof1416screenedneu-ronsintheamygdala,44neurons(3.1%)responded

in

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4837

Fig.2.Theresponsefunctionsofalltheviscosity-sensitiveneuronsto1,10,100,1000and10,000cPCMC.ThemeanandtheS.E.M.areshown.Thespontaneous ringrateforeachneuronisshownbythehorizontalline.

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

38M.Kadohisaetal./Neuroscience132(2005)33–48

Table2.NumbersofamygdalaneuronswithdifferenttypesofinputaUnimodalBimodalMultimodalOthers

Taste(G)

13G T3G T V7Temperature(T)6G V4G T F0Viscosity(V)3G F0G V F0Fat(F)

1

T V2G T V F

T F0V F

1Total231074%

5223

169

a

Note:thepercentageindicatedisofthe44orallyresponsiveneurons.

relationtooralviscosity,grittytexture,fat,taste,tempera-ture,capsaicin,LaA,and/orLiAincludedinthe25oralstimulussets.Theremainderoftheneuronswasunre-sponsivetotheoralstimuliused,exceptforfourneuronsthatincreasedtheir ringratesabovethespontaneouslevelnon-speci callytoallthestimuliapplied.(Incontrasttothe40differentialorallyresponsiveneuronsdescribedindetailhere,theyhadnosigni cantdifferentialactivityinaone-wayANOVAthattestedfordifferencesbetweenthewholesetoforalstimuli.)Ofthe40neuronswithselectiveresponsestothestimulusarray,theresponsesweretyp-icallyverysigni cant,asshownbytheone-wayANOVAacrossthe25stimuli(in35/40casesatP 0.001,in28/40casesatP 0.0001,andin25/40casesatP 10 6,andin12/40casesatP 10 10;seeExperimentalProcedures).Tocon rmthatthesigni cantresponsesofthispopulationof40neuronscouldnothavearisenbychance,weper-formedaFisher(1932;seeExperimentalProcedures)probabilitycombinationtestacrossthepopulationof1416neurons,andfounda 2valueof3664(df 2832),whichcorrespondstoazvalueof10.36,P 10 16.Thus,theresponsesofthe40neuronstotaste,viscosity,etc(acrossall25stimuli)wereveryunlikelytobeduetochance.Visual,auditory,and/orolfactoryresponseswereclearin7.4%ofthetotalsample.Neuronalresponsesrelatedtomouthmovementscomprised0.2%ofthetotalsample.Theseneuronsrespondedphasicallywheneverthemon-keymovedthemouth,andcouldbemadeto rewhenacontrolsyringecontainingnoliquidtouchedthemouthandproducedmouthmovements.Theseneuronscouldhavebeensomatosensoryormotor.Theremainderoftheneu-rons(89%)wasunresponsivetothestimuliused.Amygdalaneuronswithresponsesrelatedtotheviscosityoforalstimuli

Fig.1Ashowsaneuron(bo205)whichistunedtoviscosity,withgradedresponsestothe1–10,000cPstimulifromtheCMCviscosityseries(ANOVAwithintheviscositystimuli,(F(4,16) 7.95,P 0.0012).TheneuronalsorespondedwhentheviscositywasproducedbystimuliintheSiOseries.(Posthoctestsshowedthattheresponseswereforagivenviscositynotsigni cantlydifferenttotheCMCandthesili-cone,thoughthecorrespondenceof ringratesisnotascloseasinsomeotherbrainareas,asconsideredintheDiscussion.)Thisviscosity-sensitiveneuronalsohadtaste(F(5,22) 6.98,P 0.0007)inputs,andthedifferenceof

ringratestodifferenttemperatureswasnotsigni cant(F(3,12) 2.83,P 0.08).Thisneuronwasunusualinre-spondingtoLaA.

Thewayinwhicheachofthe17viscosity-sensitiveneuronsrespondedtothedifferentmembersoftheviscos-ityseriesisshowninFig.2,inwhichtheabscissaistheviscosityofthestimulusonalogscaleforthe veviscositystimuliintherange1–10,000cP.Forthepurposeofor-deringtheneuronsinFig.2,theneuronsweregroupedintosixsetsbasedonclusteranalysisusingtheneuronalresponsestoV1–V10,000.The rstsetofneuronsinthediagram(A–F)tendedtohaveincreasing ringratesasafunctionofviscosity.Theothersetsofneurons(G–I,J–L,MandN,OandP)weretunedwithintheviscosityseries,withincreasinganddecreasingpartsoftheirresponsefunctions.

Whilethreeoftheviscosity-sensitive(i.e.tuned)neuronsrespondedtotheSiOandtheotheroilsinwaysthatwouldbepredictediftheywererespondingtotheviscosityoftheoils,themajority(10)oftheviscosity-sensitiveneuronsrespondedtotheCMCviscosityseriesmorethantheyrespondedtotheequivalentviscositywhenprovidedbyanoil,andsome(3)didnotrespondtotheoilatall(seeTable2).AnexampleofthelatterisshowninFig.3(bo217).Theneuronwasbroadlytunedtoviscosity(V10–V1000),butits ringwhenanyoil(silicone,VO,CO,SaO;andSCwhichisanoilinwateremulsion)wasinthemouthwasatthespontaneous ringratelevel.(The ringtomostoftheaqueousstimuliwasasmallamountabovethespontaneous ringrate.)Thustheneuroncandiscriminatebetweenfattextureandviscosity(t-test,P 10 6usingalloils,andCMCV10–V1000),andthisismadeclearinFig.3B.Thisneuronalsohadrespon-sivenesstotaste(F(5,22) 3.72,P 0.014)andtemperature(F(3,13) 10.47,P 0.001),asshowninFig.3A.

Thetypicaltimecourseoftheresponsesofaviscosity-sensitiveneuronisshownintheperistimulusrastergramsandtimehistogramsinFig.4.Theneuronhadalargerresponseto10,000thanto1cPCMC.Fat-responsiveneurons

Fig.4Ashowsaneuron(bk361)whichrespondedonlytothesetofoils(withtheseresponsesbeingsigni cantlydifferentfrombothspontaneousandfromallmembersoftheviscosityseriesV10–V1000;P 0.00005).Therewasnosigni cantresponsetoanyofthemembersoftheCMCviscosityseries,nortoanyofthetastestimuli,nortoanyofthetemperaturestimuli.(TheneurondidnotrespondtoSC,whichalthoughitcontainsfat,alsocontainssweetandothertastantswhichmayhavepreventedtheneuronre-spondingtothefatinthecream.)Interestingly,theneurondidnotrespondtothefattyacidsLaAandLiA,indicatingthattheresponsestofatwerebasedonitstexture,andnotonanyfattyacidsthatmightpossiblybeproducediffatislipolyzedatallinthemouthbyanysalivarylipasethatmightbepresent.Furtherevidencethattheneuronalre-sponsewasnotbasedonfattyacidsisthattheneuronrespondedtotheSiOs(whichcontainnofatorfattyacids,buthaveasimilartexturetothefattyoilssuchasVO,COandSaO).Fig.4Billustratestheneuronalresponsesin

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4839

Fig.3.(A)Theresponsesofanamygdalaneuron(bo217)withdifferentialresponsestotaste,temperatureandviscosity.Theneurondidnotrespondtofattexture.ConventionsasinFig.1.G,N,M,HandQarethetastestimuli.T10–T42arethetemperaturestimuli.V1–V10,000aretheCMCviscosityserieswiththeviscosityincP.ThefattexturestimuliwereSiO10,SiO100,SiO1000(SiOwiththeviscosityindicated),VO,COandSaO.BJisfruitjuice;Capis10 Mcapsaicin;LaAis0.1mMLaA;LiAis0.1mMLiA;seeTable1.(B)The ringrate( S.E.M.)todifferentviscositiesofCMCshownasagraph.Thehorizontallineshowsthespontaneousactivity( S.E.M.).Therewasnoresponsetothefattyoils.AbbreviationsasinFig.1.

moredetail,byshowingperistimulustime ringratehisto-gramsandrastergramsforsomeofthestimuli,togetherwithaninsettoshowthespikebeingrecorded,whichwas,aswasusual,verywellisolated.

AsshowninTable2,twooftheneurons(bk361andbk364)werefatsensitive,inthattheirresponseswerelargetotheoilsandoccurredinawaythatwouldnotbepredictedfromanysmallerresponsetheymighthavetotheCMCviscosityseries(seeexampleinFig.4A).BothoftheseneuronsalsorespondedtotheSiO,andneitherofthemrespondedtofattyacids.Inaddition,noneofnineotherneuronswitharesponsetooneorbothofthefattyacidswasclassi edasfatsensitive.Temperature-responsiveneurons

Fig.1Bshowsaneuron(bo219c2)withdifferentialresponsestodifferenttemperatures(F(3,16) 9.179,P 0.001).TheneuronrespondedprimarilytoT10fromthetemperatureseries(andhadnosigni cantlydifferentresponseswithinthetasteseries,withintheviscosityseries,orwithinthefatsoroils).

Thepro lesoftheresponsivenesstothedifferenttem-peraturestimuliofthethermosensitiveneuronsareshowninFig.5.TheorderoftheneuronsinFig.5justforthepurposesofillustrationisbasedonsixclustersidenti edbyclusteranalysis.Inthe rstset(A–G),theneuronalresponsesshowagenerallyupwardtrendwithincreasingtemperature(T10–42);inthesecondset(H–L),theneu-ronshadthebestresponsestoT10;inthethirdset(M–P),theneuronsshowedmoreactivitytothelowthantothehightemperatures.Theotherneurons(QandRinFig.5)wereseparateclusters.

Ofthe18temperature-sensitiveneurons,sixrespondedonlytotemperature(seeTable2).Thusthetemperature

of

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

40M.Kadohisaetal./Neuroscience132(2005)33–48

Fig.4.(A)Responsesofanamygdalaneuron(bk361)respondingtofat.ConventionsasinFig.1.G,N,M,HandQarethetastestimuli.T10–T42arethetemperaturestimuli.V1–V10,000aretheCMCviscosityserieswiththeviscosityincP.ThefattexturestimuliwereSiO10,SiO100,SiO1000(SiOwiththeviscosityindicated),VO,COandSaO.BJisfruitjuice;Capis10 Mcapsaicin;LaAis0.1mMLaA;LiAis0.1mMLiA;seeTable1.(B)Post-stimulus-timehistogramandrastergramofthesameneuron(bk361)toshowthenatureoftheneuronalresponsesontypicaltrialstoeachstimulus.Spont,spontaneous ringrate.Therecordingoneachtrialstartedattime0,whenthestimuluswasdelivered.ThepoststimulustimehistogramswereGaussiansmoothedwithaS.D.ofonetimebin,each100mswide.Theinsetshowsspikesfromtheneuronbeingrecordedtoshowthetypicallyverygoodisolationofthespikes,andthereisnooverlapwiththenoise.

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4841

Fig.5.Theresponsefunctionsofallthethermosensitiveneuronstodifferenttemperaturein°C(T10,T23,T37andT42).ThemeanandtheS.E.M.areshown.Thespontaneous ringrateforeachneuronisshownbythehorizontalline.

whatisinthemouthisrepresentedindependentlyoftaste,viscosity,andfattytextureintheprimateamygdala.Inaddi-tion,10neuronsrespondedtobothtasteandtemperature,showingthattheprimateamygdalarepresents

combinations

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

42M.Kadohisaetal./Neuroscience132(2005)33–48

Fig.6.Astimulusspace(multidimensionalscaling)ofthestimulussimilaritybasedontheacross-neuronresponsepro lesofthe44orallyresponsiveamygdalaneurons.Thetastestimuliwere1Mglucose(G),0.1MNaCl(N),0.1MMSG(M),0.01MHCl(H)and0.001MQuinine–HCl(Q);thetemperaturestimuliwereT10,T23,T37andT42wherethenumberindicatesthetemperaturein°C;theviscositystimuliwereV1,V10,V100,V1000andV10,000wherethenumeralindicatestheviscosityincP);fattexturestimuliwereSiO10,SiO100,SiO1000(SiOwiththeviscosityindicated),VO,COandSaO.BJisfruitjuice;Capis10 Mcapsaicin;LaAis0.1mMLaA;LiAis0.1mMLiA.Thesolidlinejoinsthemembersoftheviscosityseries.Differentlinestylesjointhemembersofthetaste,temperatureandoilstimuli.Thetwo-dimensionalsolutionaccountedfor91%ofthevariance(n 44).

ofthesetwomodalities,potentiallyprovidingthebasisfordifferentbehavioralresponsestoparticularcombinationsofthetasteandtemperatureofthefoodor uidinthemouth.AsshowninTable2,someneuronsintheprimateamygdalarespondtooraltemperatureandtoviscosity,andinadditionotherneuronsrespondtotemperaturecombinedwithseveralothertypesoforalsensorystimulusincludingtasteandviscosity.

Populationanalyses

Therepresentationofthesimilarityofthestimulibythepop-ulationofneuronswasapproachedwithmultidimensionalscalinganalysis,basedonthe rst1sofpost-stimulusactiv-ity,andwasperformedontheresponsesofthesame44neurons(Fig.6).Thesegregationsbetweenmodalitiesareclearlyshowninthemultidimensionalspace.Thedifferentmodalitieshavebeenjoinedbylinestohelpclarifytherep-resentationinthismultidimensionalspace.First,theviscosityseriesisverywellseparatedinthespace(primarilyalongthexaxis).The vetastestimuliarewellseparatedfromeachother,butcontainedintheirownpartofthespaceseparatefromalltheotherstimuli.Thetemperatureseriesareagainclearlylaidoutinthespace(primarilyalongtheyaxis).Theoilsarelocatedcloselytogetherandclearlyseparatefromtheviscosityseriesparametricrepresentation.Itisofconsider-ableinterestthattheoilstimuliarenotseparatedoutinthespaceaccordingtotheirviscosity,asthisprovidesfurtherevidencethattheviscosityofstimuliisencodedparametri-callyintheamygdala;andthatfattytextureiscodedasafattytextureindependentlyofitsviscosity.Inacheckthatthemultidimensionalspacerepresentedthedistancesbetweenthestimulireasonablyandinawaythatcouldbeinterpreted,wefoundthatremovingdifferentrandomlypermutedsetsof veneuronsfromthecalculationsusedforthemultidimen-sionalspacedidnotin uencewhatisshowninFig.6.

Ofthe44neuronsinthesample,23(52%)neuronswereunimodal(13unimodaltaste,sixunimodaltempera-ture,threeunimodalviscosityandoneunimodalfatneu-rons),10(23%)neuronswerebimodal,andseven(16%)neuronsweremultimodalwithresponsestotaste,temper-atureandviscosity(seeTable2;thefourneuronsshownas‘other’hadsigni cantlydifferentresponsestotheset

of

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4843

oralstimulibasedontheANOVAperformedacrossthewholesetoforalstimuli,butwerenotfurtherclassi edonthebasisoffurtherANOVAsastemperaturesensitiveetc).The ndingsprovideclearevidenceforconvergenceoftasteandsomatosensory(thermosensitive,texture-sensitiveand/orfatsensitive)inputsontosomeneuronsintheamygdala(seeFigs.1Aand3),andalsothateachtypeofinputisrepresentedindependentlyoftheothers(seeFigs.4and1B).Further,someneuronshadresponsive-nesstoLaA,LiAandCapwhencomparedwiththeirsol-ventwater(T23/V1):fourneuronstoLaA,sevenneuronstoLiA(withtwooftheseneuronsrespondingtobothLaAandLiA);andfourneuronstoCap.Noneofthefattyacidsensitiveneuronswereclassi edasfatresponsive.Inaddition,threeneuronsrespondedtotheGr(whencom-paredwiththeequallyviscousbutnotgrittyV1000).

Thebreadth-of-tuningmetric(SmithandTravers,1979)calculatedacrossthetastestimuliH,Q,NandGwas0.85 0.03(mean S.E.M.)forthe(13)neuronswithonlytasteinputs(i.e.withoutsomatosensory-thermosensitive,texture-sensitiveand/orfatsensitive-input),andfor(14)neu-ronswithbothtasteandsomatosensoryinputswas0.93 0.02(P 0.07).Thecorrespondingsparsenesseswere0.77 0.04and0.87 0.04(P 0.092).Inaddition,themeansparsenessoftherepresentationof16stimuli(G,BJ,N,M,H,Q,T23/V1,T10,T37,T42,V10,V100,V1000,SCandVO)ofthe44amygdalaneuronswas0.79 0.18(mean S.D.).Thiscomparestothemeansparsenessof52orbitofrontalcortexneuronstothesamesetofstimuli(Verhagenetal.,2003)of0.67 0.23(mean S.D.;P 0.006),whichindicatestheamygdalaneuronsweretunedmorebroadlytothesetofstimuli.

Localizationofrecordings

ThereconstructedpositionsoftheneuronsanalyzedinthisstudyareshowninFig.7.Thetoprowshowsthelocationsoftheneuronsrespondingtooralstimuli.Mostoftheseneuronswerelocatedinthebasal/basalaccessoryamygdaloidnu-cleus,centralamygdaloidnucleusandlateralamygdaloidnucleus.Althoughitwasnotanaimofthisstudytoinvesti-gateamygdalaneuronswithotherresponses,wedid ndanumberofneuronswithvisualresponses(93;categorizedassuchbythecriteriaofSangheraetal.,1979;Leonardetal.,1985),withauditoryresponses(11),witholfactoryresponses(2),orwithactivityassociatedwithmouthmovements(4),asshowninthemiddlerowofFig.7,primarilyforcomparisonwiththelocationsoftheorallyresponsiveneurons.Someoftheseneuronswithotherthanoralresponsesrespondedtostimuliinseveralmodalities,asshowninFig.7.Eightofthe44neuronswithoralresponsesanalyzedhereandshowninTable2hadvisual(6)orauditory(2)responses.Thebottomrowshowsthesitesofalltherecordedneurons,toindicatethepartsoftheamygdalathatweresampled.

DISCUSSION

Theresultsinthispaperdescribethediscoveriesofneu-ronsintheprimateamygdalawithactivityrelatedtotheviscosity,thefattexture,andthetemperatureoffoodinthe

mouth.Theresultsalsodescribethediscoveryofamyg-dalaneuronstunedtocapsaicin,oralgrittiness,andtofattyacids.Theseresultsalsoshowthattheprimateamygdalacontainsseparaterepresentationsofthepropertiesoforalstimuli,butalsohasotherneuronsthatcombinethemwitheachother,andalsowithtaste.These ndingsprovidethe rstdetailedevidenceontherepresentationintheprimateamygdalaofnon-tasteoralstimuli,andarerelevanttounderstandingwhatisrepresentedinthehumanamyg-dala.Suchstimuli,sometimescombinedwithtaste,pro-videimportant(mainlyprimaryorunlearned)reinforcers(rewardsandpunishments)forfoodintake,andinthiswayareinvolvedinthecontrolofappetiteandthecontroloffoodintake.Thesestimulialsoprovidetheprimaryrein-forcerformuchstimulus-reinforcerassociationlearningwithappetitivestimuli,inwhichtheamygdalaisimplicated(EverittandRobbins,1992;LeDoux,1995,2000;GallagherandChiba,1996;RobbinsandEveritt,1996;Rolls,1999,2000;Schoenbaumetal.,1999;BaxterandMurray,2000;DavisandWhalen,2001;Everittetal.,2003;PetrovichandGallagher,2003).Thisstimulus-reinforcerassociativelearningallowspreviouslyneutralvisual,olfac-tory,andauditorystimulitobecomeassociatedwiththesensorypropertiesoffoodsinthemouth.Bythisassocia-tivelearning,suchpreviouslyneutralstimulicangaincon-trolofbehavior.Forexample,intheratthecentralnucleioftheamygdalaencodeorexpressPavlovianstimulus-response(CS-UR)associations(includingconditionedsuppression,conditionedorienting,conditionedautonomicandendocrineresponses,andPavlovian-instrumentaltransfer);andmodulate,perhapsbyarousal,theassocia-bilityofrepresentationsstoredelsewhereinthebrain(GallagherandHolland,1994;HollandandGallagher,1999).Incontrast,thebasolateralamygdalaencodesorretrievestheaffectivevalueofthepredictedUS,andcanusethistoin uenceaction–outcomelearningviapath-waystobrainregionssuchasthenucleusaccumbensandprefrontalcortexincludingtheorbitofrontalcortex(Cardinaletal.,2002).Forbothtypesoflearningintheamygdala,thenatureoftheprimaryreinforcerisimportant,forthisde neswhatthelearningisabout,andinthecaseofappetitiveoralreinforcers,thetypesofneurondescribedinthispaperarefundamental.

Therepresentationofviscositydescribedhereen-codesthedegreeofviscosityofwhatisinthemouth,inthateachneuronhasgraded ringtothedifferentviscos-itiesused(CMCintherange1–10,000cP),andinthatdifferentneuronshavedifferentresponsefunctions,asshowninFig.2.FurtherevidenceforthisisprovidedbythemultidimensionalspaceshowninFig.6,inwhichthedif-ferentviscositystimuliareparametricallyrepresentedandwellseparatedfromeachotherinthestimulusspace.Thehard,round,microspheresweemployed(100–300 m)evokeanoralgrittytexture,andthiswasaneffectivestimuluswhensuspendedincelluloseforthreeneurons(whencomparedwithequallyviscouscellulose).

Fatinthemouthwasrepresentedintwowaysbytheamygdalaneuronsdescribedhere.Onewaywasbytheamygdalaneuronsthatrespondtofatandnottothecellulose

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

44M.Kadohisaetal./Neuroscience132(2005)33–48

Fig.7.Thereconstructedpositionsinthebrainoftheneuronsinthisstudy.Top:thesymbolwithwhichthelocationofeachneuronisindicatedshowswhethertheneuronwastunedtog taste,v viscosity,t temperature,f fatortocombinationsofthese.Middle:thesymbolwithwhichthelocationofeachneuronisindicatedshowswhethertheneuronwastunedtovis visual,aud auditory,olf olfactoryortocombinationsofthese,ortomouthmovement.Bottom:thesitesoftheneuronsthatwerenotdifferentiallyresponsivetothestimuliusedinthisinvestigation,toshowtheareaofamygdala

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4845

Fig.8.Afrequencyhistogramshowingthecorrelationsbetweenthe ringofeachofthe44orallyresponsiveneuronsandtheacceptabilityratings.Anormaldistributionhasbeen tted.Themeancorrelationwas0.09andtheS.D.was0.22.

viscosityseries(Fig.4).Theseneuronsencodefatbyitstexture(andnotbyanyodororfreefattyacidcue),inthatthesameneuronsrespondtoSiO,toMO,andnottofattyacids(Gilbertson,1998;Verhagenetal.,2003).Thesecondwayinwhichfatisdistinguishedfromnon-fattexturesintheamyg-dalaisbytheneuronsthatrespondtoviscosityandnottotheoils(seeexampleshowninFig.3).Indeed,itwasofinterestthatmostoftheneuronstunedtothecelluloseviscosityseries(13/17)tendedtohavesmallerresponsestothesameviscositywhenproducedbyfat,providingafurtherwayinwhichthepopulationofamygdalaneuronsdescribedhereseparatestherepresentationsoforalviscosityandfat.Inaddition,thefewneuronsthatrespondedtofattyacidsdidnotrespondtotheoilstimuli.Wenotethattheresponsesoftheseneuronstotheoils(Siseries)orthetexture(carboxymethylcellulose)stimulicannotbeaccountedforbytastere-sponses,asshownbythefactsthat7/18amygdalaneuronsthatrespondedtothesetexturestimulihadnoresponsestothetastestimuliG,N,H,QandM;thatneuronsthatdidrespondtotasteandtexturestimulihaddifferentresponsive-nesstoeachothertothetasteandtexturestimuli(seeexamplesinFigs.1and3);andthathumanpsychophysicalratingona100mmvisualanalogratingscaleoftheintensityofthetasteofthetexturestimuliwas12.5 3.6(S.D.),andofthewaterwas11.6 3.5,whereasthatforthetastestimuliwas61.0 10.0.(Consistentwiththis,carboxymethylcellu-loseisusedbyindustryasatastelessandodorlessthicken-ingagent.)Further,theresponsesoftheseamygdalatextureneuronscannotbeaccountedforbyresponsivenesstoodor,inthatthehumanpsychophysicalratingonthe100mmvisual

analogratingscaleoftheintensityoftheodoroftheCMCtextureandoil(SiOseries)stimuliwas8.6 2.7(mean S.D.),andofthewaterwas7.8 2.6.

Therepresentationoftemperatureprovidedbytheseprimateamygdalaneuronswasgraded,asshownbytheresponsesoftheneuronsillustratedinFig.1BandFig.5,andbythemultidimensionalspaceshowninFig.6inwhichthewarmtemperaturesT42,T37areparametricallysepa-ratedfromtheotherstimuliincludingT23/T10.Fourofthe44amygdalaorallyresponsiveneuronstestedinthisstudyhadresponsestocapsaicinthatweredifferentfromwater.Theneuronsdidnotrespondto42°Cwater,andthismayberelatedtothefactthatthesensationofcapsaicinismediatedbythevanilloidreceptorsubtype1,whichre-spondstotemperaturesabove43°C(Caterinaetal.,1999).

Theresponsesoftheviscosity,fattexture,andtem-peratureresponsiveneuronswerenotrelatedtoanymouthmovementsasshownbythefollowing.First,Fig.4showsrastergramsandperistimulusresponsehistogramsforaneuronthatrespondedtoalltheoils,andnophasicresponsivenesswasfoundtothesestimulithatmighthavebeenrelatedtoanymouthmovement.Moreover,theneu-rondidnotrespondtoanyoftheCMCseries,towhichmouthmovementswerealsomade.Further,theseamyg-dalaneuronsintheirresponsivenesstotexturedstimuliwereverysimilartotheresponsesofneuronsintheor-bitofrontalcortexandinsulartastecortexthatrespondtoviscosityandfattexture,whichalso,asillustratedinthefollowingpapers,donothavephasicresponsivenessre-latedtomouthmovements(Rollsetal.,2003;Verhagenetal.,2003,2004).Moreover,thetuningfunctionsofdif-ferentamygdalaneuronstothesetofviscosity,oil,andtemperaturestimuliwereverydifferenttoeachother,sothatnosinglefactorsuchasmouthmovementscouldeasilyaccountfortheresponses.Further,neuronalre-sponsesrelatedtomouthmovementswerefoundandcomprised0.2%ofthetotalsample.Theseneuronsre-spondedphasicallywheneverthemonkeymovedthemouth,andcouldbemadeto rewhenacontrolsyringecontainingnoliquidtouchedthemouthandproducedmouthmovements.Theseneuronsintheirphasic,mouth-movementrelated,responseswereverydifferentfromthoseofthetaste,texture,oilandtemperaturesensitiveneurons.

Someoftheamygdalaneuronsdescribedhereprovideseparaterepresentationsofviscosity,fattexture,temper-ature,taste,capsaicin,grittiness,andfattyacids,andotherneuronscombinedinputsfromdifferentsubsetsofthesepropertiesofsensorystimuli(seeTable2).Thecombina-tion-respondingneuronsprovideonepossiblebasisfordifferentbehavioraland/orendocrineresponsestopartic-ularcombinationsofthesensorypropertiesofstimulisuch

inwhichneuronsweresampled.1–2Pshowsthatthecoronalsectionwastaken1–2mmposteriortothesphenoidprocessusedasalandmark,whichisatapproximatelytheA-Pleveloftheopticchiasm.ThearchitectonicboundariesasdescribedbyAmaraletal.(1992)areindicated.AB,accessorybasalamygdaloidnucleus;acp,anteriorcommissureposteriorpart;B,basalamygdaloidnucleus;Ce,centralamygdaloidnucleus;Cl,claustrum;L,lateralamygdaloidnucleus;rh,rhinal ssure;sts,superiortemporal

sulcus.

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

46M.Kadohisaetal./Neuroscience132(2005)33–48

asfoodinthemouth.Otherbrainareasthatalsoprovidepossiblebasesforparticularbehavioralresponsestodif-ferentcombinationsoforalstimuliincludetheorbitofrontalcortex(Rollsetal.,2003).Thesebehaviorsincludesensory-speci csatiety,inwhichthepleasantnessofafoodeateninamealdecreasesrelativetootherfoodsnoteateninthemeal(Rollsetal.,1981;Rolls,1997,1999,2004;Kringelbachetal.,2003).Withrespecttothepleas-antnessofthesensorypropertiesoffood,thereisnowverystrongevidencethatorbitofrontalcortexneuronsarere-latedtosensory-speci csatiety(Rollsetal.,1989;CritchleyandRolls,1996;Kringelbachetal.,2003),whereasthereislessevidenceforstrongmodulationbyhungerorhedonicsoftasteresponsivenessintheamygdala(Nishijoetal.,1988a,b;YanandScott,1996;RollsandScott,2003).Thefactthatsomeamygdalaneuronsrespondtobothtasteandtemperatureshowsthatthetemperatureofwhatisinthemouthisnotencodedonlyseparatelyfromtheothersensorypropertiesofthefood,butalsoincombinationwithothersensorypropertiesoffood.Thusthistemperaturerepresentationmaynotonlyallowhotorcoldsubstancestoberejected(oraccepted),butalsoenablefoodsthathaveparticularcombinationsoftemperature,tasteandtexturetobereactedtodifferently.

WenotethatalthoughhumanfMRIresultsareconsis-tentwiththosedescribedhereinshowingforexamplethattheorbitofrontalcortex,whichhasmajorreciprocalcon-nectionswiththeamygdala(Amaraletal.,1992),isstronglyactivatedbytexturedwholefoodstimulisuchastomatojuiceandchocolate(Kringelbachetal.,2003),andhasarepresentationoforalviscosity(deAraujoandRolls,2004).However,thedetailsoftherepresentationasde-scribedhere,withbothunimodalneurons,andbimodalneuronsshowingconvergence,togetherwiththedetailsoftheindividualneuronaltuningtoviscosityandtemperaturestimuli,andtheseparatenessoftherepresentationfromgrittyandcapsaicin,couldnotbeshownbyfMRIstudies.Inaddition,fMRIstudiesprovidelittleevidenceontheproportionofneuronsinastructurethatrespondindiffer-entways.Oneratherinterestingresultofwhatwasfoundintheprimateamygdalaisthatprovidedthatcareistaken,aswedid,toexcludemovement-relatedneuronsfromthesample,thenthenumberofneuronsspeci callytunedtooralsensorystimuliintheamygdalaappearsquitemodest,3.1%.Eventhoughtheproportionisnotlarge,thepopu-lationofneuronsdescribedheredidprovideverydetailedrepresentationsofwhatisinthemouth,andindeed,ifscaledupaccordingtothenumberofneuronsintheex-tensivepartoftheamygdalasampled(seeFig.7),wouldamounttothousandsofneurons,whichissuf cienttoprovideagreatdealofinformationaboutwhichstimulusispresentinthemouth(RollsandTreves,1998;RollsandDeco,2002).Indeed,informationofthistype,thepropor-tionofneurons,andhoweachoneresponds,isofcourseessentialtounderstandinghowthebrainoperates,andtheimportanceoftheseneuronstooverallbehaviorisattestedtobytheevidencethattheamygdalaisinvolvedinasso-ciativelearningforwhichoneimportanttypeofuncondi-tionedstimulusisfoodinthemouth(Aggleton,2000).Thefactthattheproportionofamygdalaneuronsdevotedtooralstimuliisnothighisconsistentwiththefactthatotherinputs,includingvisual(Sangheraetal.,1979;Leonardetal.,1985;Rolls,2000),arerepresentedintheprimateamygdala,andiftheoverallrepresentationistobemain-tainedsparse,whichhasmanyadvantagesforneuralcomputation(OlshausenandField,1996;RollsandDeco,2002),thentheproportiondevotedtoanyonemodalitycannotbehigh.Moreover,thepreciseidenti cationde-scribedhereoftheproportionsofneuronsactivatedbyoralstimuliintheamygdalaisofinterestinrelationtofunctionalmagneticresonanceneuroimaging,inwhichactivationstothetasteofglucoseorsaltcanbedemonstratedinthehumanamygdala.Thisisanindicationthatevenquitelowproportionsofneuronsthatareactivatedmaybesuf cienttoproduceaverysigni cantfMRIsignal,atleastina3Tscanner(O’Dohertyetal.,2001).

Inthispapertherewardvalueofthetexturedstimulihasnotbeenmanipulated,byforexamplefeedingthemonkeytosatietytodecreasetherewardvalueofthestimulus,andthenre-measuringtheneuronalresponse.Ithaspreviouslybeenshownthatsatietyproducesarathermodest(onaverage58%)reductionintheresponsesofmacaqueamygdalaneu-ronstotaste(YanandScott,1996),incomparisontotheessentiallycompletereductionofresponsivenessfoundinorbitofrontalcortextasteneurons(Rollsetal.,1989).Further,therepresentationintheamygdalaoftheseoralstimulidoesnotappeartobeonanysimplehedonicbasis,inthatnodirectioninthemultidimensionalspaceinFig.6re ectsthemeasuredpreferenceofthemonkeysforthestimuli.More-over,foreachmonkey,therewasnocorrelationbetweentheacceptabilityratingsofeachstimulusbythemonkeysandtheaverageneuronalresponsestoeachstimulus(BK:F(1,21) 0.33,P 0.570;BO:F(1,23) 1.85,P 0.187).Wealsocalculatedacorrelationbetweentheacceptabilityratingsofthemonkeyandtheresponsesofeachneuronthathadasigni cantresponsetotheoralstimuli.ThevaluesofthecorrelationsareshowninFig.8,togetherwitha ttednormaldistribution.For39ofthe44neuronstherewasnosigni cantcorrelationbetweentheneuronalresponseandtheprefer-enceratingacrossthe25stimuli.For veneurons,therewasasigni cantcorrelationatP 0.05,andfortheseneuronsthecorrelationvalueswere0.40(P 0.05),0.42(P 0.05),0.46(P 0.05),0.57(P 0.003)and0.58(P 0.002).Acrossall44neurons,themeancorrelationwas0.08withastandarddeviationof0.22(asshowninFig.8).(Thisanalysistakesintoaccountthattheacceptabilitymeasureforeachstimuluswasdifferentforsomeofthestimuli,asshowninTable1.)Afurtheranalysisshowedthatthelargestdifferenceswerethatthecarboxymethylcellulose(V10,000)stimuluswaslessac-ceptableinbk,theSCwaslessacceptableinbo;andthegrittystimuluswaslowerinbk.Nevertheless,the ringratesofamygdalaneuronstothesethreeparticularstimulishowednodifferencebetweenthemonkeys,P 0.43,0.55and0.80,respectivelybetweenboandbkacrossalltheiramygdalaneurons;t-test.Moreover,evenwhenthe ringratestothesethreestimuliwerescaledineachmonkeyrelativetothe ringratestotheacceptableglucoseandunpleasantsaline,thescaledvaluesforthethreestimuliwerenotsigni cantlydif-

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

M.Kadohisaetal./Neuroscience132(2005)33–4847

ferentforthetwomonkeys(P 0.33forallthreecompari-sons)andwerenotrelatedtothemonkeys’preferences(P 0.12)forthesethreestimuli.Thusweconcludebythesedistincttypesofanalysis,thatthedifferencesinthe ringratesoftheseneuronstothesetofstimuliinmostcasesdidnotre ectjusttheacceptabilitytothemonkeyofthestimuli,butre ectattributesofthestimulisuchastheirtexture,taste,andtemperature.Onlytwoneuronshadreasonablysigni -cantcorrelationsoftheir ringrateswiththeacceptabilityratings(giventhat44correlationsweretested,andamorestringentcriterionthanP 0.05shouldbeapplied),andevenforthesetwoneuronsthecorrelationswereonlymoderate,0.57and0.58.

Incomparisontotheorbitofrontalcortexneuronsthatrespondtothesamesetofstimuli(Rollsetal.,2003;Verhagenetal.,2003;Kadohisaetal.,2004),theamyg-dalaneuronshadless netuningacrossthesetof16stimuliusedinbothareas,asindicatedbythesparsenessmeasurespresentedhere.Interestingly,theamygdalaneuronswithoralresponsivenesshadveryclearlysepa-ratedresponsestotheviscosityseriesandthetempera-tureseriesinthemultidimensionalspace,witharelativelycompactpartofthespacedevotedtotaste(seeFig.6).Incontrast,themultidimensionalspacefor53orbitofrontalcortexneuronstestedwiththesamesetofstimuli(apartfromminordifferencesintheoilystimulussubset)inthesametwomonkeysshowedarelativelygreaterareade-votedtotaste,andrelativelymorecompressedrepresen-tationsofviscosityandtemperature(Rollsetal.,2003;Verhagenetal.,2003;Kadohisaetal.,2004).Thismaybeconsistentwiththemajorgustatoryinputstotheorbitofron-talcortexfromtheprimarytastecortexintheinsular/opercularregions(Baylisetal.,1994),andwiththemajorsomatosensoryinputstotheamygdalafromsomatosen-soryparts(Friedmanetal.,1986)andalsofromtasteparts(Verhagenetal.,2004)oftheinsula.Afurtherinterestingdifferenceisthatintermsofbestresponsestodifferenttastes,57%oftheorbitofrontalcortextasteneuronshadtheirbestresponsestoglucose,whereas21%oftheamygdalaneuronshadtheirbestresponsetoglucose( 2 12.5,df 5,P 0.03).(MoreamygdalaneuronshadtheirbestresponsestoH(18%)andM(14%).)

Inconclusion,weshowforthe rsttimethatamygdalaneuronscanbespeci callytunedto vetypesoforalsomatosensorystimulus:viscosity,grittiness,fattexture,capsaicin,andtemperature.Inadditiontoseparaterepre-sentationsprovidedbysomeneurons,otherneuronsre-spondtocombinationsofviscosityand/ortasteand/orgrittyand/orcapsaicinand/orfatinputs,therebyprovidingarichrepresentationofthesensorypropertiesoffoodinthemouth,inwhichparticularcombinationsoftheabovepropertiescanberepresentedseparatelyfromthecompo-nents.Thisallowsforsubjectiveandbehavioralresponsesthatcanbebasedonparticularcombinationsoftheseinputs,providingforgreatsensorycapabilityinchoosingandlearningaboutparticularcomplexfoods,andinex-tendingthe avorspace.

Acknowledgments—ThisresearchwassupportedbyMedicalRe-searchCouncilgrantPG9826105toE.T.Rolls.SamanthaLinetookpartinsomeoftherecordingsaspartofagraduateresearchproject.

REFERENCES

AggletonJP,ed(2000)Theamygdala.Oxford:OxfordUniversityPress.AmaralDG,PriceJL,PitkanenA,CarmichaelST(1992)Anatomical

organizationoftheprimateamygdaloidcomplex.In:Theamygdala(AggletonJP,ed),pp1–66.NewYork:Wiley-Liss.

BaxterMG,MurrayEA(2000)Reinterpretingthebehaviouraleffectsof

amygdalalesionsinnon-humanprimates.In:Theamygdala:afunc-tionalanalysis(AggletonJP,ed),pp545–568.Oxford:OxfordUni-versityPress.

BaylisLL,RollsET,BaylisGC(1994)Afferentconnectionsofthe

orbitofrontalcortextasteareaoftheprimate.Neuroscience64:801–812.

BerthoudHR(2003)Neuralsystemcontrollingfoodintakeandenergy

balanceinthemodernworld.CurrOpinClinNutrMetabCare6:615–620.

BourneMC(2002)Foodtextureandviscosity:conceptandmeasure-ment.London:AcademicPress.

CardinalN,ParkinsonJA,HallJ,EverittBJ(2002)Emotionand

motivation:theroleoftheamygdala,ventralstriatum,andprefron-talcortex.NeurosciBiobehavRev26:321–352.

CaterinaMJ,RosenTA,TominagaM,BrakeAJ,JuliusD(1999)A

capsaicin-receptorhomologuewithahighthresholdfornoxiousheat.Nature398:436–441.

CritchleyHD,RollsET(1996)Hungerandsatietymodifythere-sponsesofolfactoryandvisualneuronsintheprimateorbitofrontalcortex.JNeurophysiol75:1673–1686.

DavisM(1994)Theroleoftheamygdalainemotionallearning.IntRev

Neurobiol36:225–266.

DavisM,WhalenPJ(2001)Theamygdala:vigilanceandemotion.Mol

Psychiatry6:13–34.

deAraujoIET,RollsET(2004)Therepresentationinthehumanbrain

offoodtextureandoralfat.JNeurosci24:3086–3093.

EverittB,CardinalRN,ParkinsonJA,RobbinsTW(2003)Impactof

amygdala-dependentmechanismsofemotionallearning.AnnNYAcadSci985:233–250.

EverittB,RobbinsTW(1992)Amygdala-ventralstriatalinteractions

andreward-relatedprocesses.In:Theamygdala(AggletonJP,ed),pp401–430.Chichester:Wiley.

FeigenbaumJD,RollsET(1991)Allocentricandegocentricspatial

informationprocessinginthehippocampalformationofthebehav-ingprimate.Psychobiology19:21–40.

FisherRA(1932)Statisticalmethodsforresearchworkers.London:

OliverandBoyd.

FrancisS,RollsET,BowtellR,McGloneF,O’DohertyJ,BrowningA,

ClareS,SmithE(1999)Therepresentationofthepleasanttouchinthebrainanditsrelationshipwithtasteandolfactoryareas.Neuroreport10:453–459.

FriedmanDP,MurrayEA,O’NeillJB,MishkinM(1986)Corticalcon-nectionsofthesomatosensory eldsofthelateralsulcusofmacaques:evidenceforacorticolimbicpathwayfortouch.JCompNeurol252:323–347.

GallagherM,ChibaAA(1996)Theamygdalaandemotion.CurrOpin

Neurobiol6:221–227.

GallagherM,HollandPC(1994)Theamygdalacomplex:multipleroles

inassociativelearningandattention.ProcNatlAcadSciUSA91:11771–11776.

GilbertsonTA(1998)Gustatorymechanismsforthedetectionoffat.

CurrOpinNeurobiol8:447–452.

HollandPC,GallagherM(1999)Amygdalacircuitryinattentionaland

representationalprocesses.TrendsCognSci3:65–73.

KadohisaM,RollsET,VerhagenJV(2004)Orbitofrontalcortexneu-ronalrepresentationoftemperatureandcapsaicininthemouth.Neuroscience127:207–221.

Abstract—The primate amygdala is implicated in the control of behavioral responses to foods and in stimulus-reinforcement learning, but only its taste representation of oral stimuli has been investigated previously. Of 1416 macaque amygdala neurons record

48M.Kadohisaetal./Neuroscience132(2005)33–48

RollsET,ScottTR(2003)Centraltasteanatomyandneurophysiology.

In:Handbookofolfactionandgustation(DotyRL,ed),pp679–705.NewYork:Dekker.

RollsET,SienkiewiczZJ,YaxleyS(1989)Hungermodulatesthe

responsestogustatorystimuliofsingleneuronsinthecaudolateralorbitofrontalcortexofthemacaquemonkey.EurJNeurosci1:53–60.

RollsET,ToveeMJ(1995)Sparsenessoftheneuronalrepresentation

ofstimuliintheprimatetemporalvisualcortex.JNeurophysiol73:713–726.

RollsET,TrevesA(1998)Neuralnetworksandbrainfunction.Oxford:

OxfordUniversityPress.

RollsET,VerhagenJV,KadohisaM(2003)Representationsofthe

textureoffoodintheprimateorbitofrontalcortex:neuronsrespond-ingtoviscosity,grittinessandcapsaicin.JNeurophysiol90:3711–3724.

RollsET,YaxleyS,SienkiewiczZJ(1990)Gustatoryresponsesof

singleneuronsinthecaudolateralorbitofrontalcortexofthema-caquemonkey.JNeurophysiol64:1055–1066.

SangheraMK,RollsET,Roper-HallA(1979)Visualresponsesof

neuronsinthedorsolateralamygdalaofthealertmonkey.ExpNeurol63:610–626.

SchoenbaumG,ChibaAA,GallagherM(1999)Neuralencodingin

orbitofrontalcortexandbasolateralamygdaladuringolfactorydis-criminationlearning.JNeurosci19:1976–1884.

ScottTR,KaradiZ,OomuraY,NishinoH,Plata-SalamanCR,LenardL,

GizaBK,AoS(1993)Gustatoryneuralcodingintheamygdalaofthealertmonkey.JNeurophysiol69:1810–1820.

ScottTR,YaxleyS,SienkiewiczZJ,RollsET(1986a)Gustatory

responsesinthefrontalopercularcortexofthealertcynomolgusmonkey.JNeurophysiol56:876–890.

ScottTR,YaxleyS,SienkiewiczZJ,RollsET(1986b)Tasteresponses

inthenucleustractussolitariusofthebehavingmonkey.JNeuro-physiol55:182–200.

SmithDV,TraversJB(1979)Ametricforthebreadthoftuningof

gustatoryneurons.ChemSenses4:215–219.

SteinbergerJ,DanielsSR(2003)Obesity,insulinresistance,diabetes,

andcardiovascularriskinchildren.Circulation107:1448–1453.SzolcsanyiJ(1990)Capsaicin,irritation,anddesensitisation:neurophysio-logicalbasisandfutureperspective.ChemSenses2:141–168.

VerhagenJV,KadohisaM,RollsET(2004)Theprimateinsular/

operculartastecortex:neuronalrepresentationsoftheviscosity,fattexture,grittinessandtasteoffoodsinthemouth.JNeuro-physiol92:1685–1699.

VerhagenJV,RollsET,KadohisaM(2003)Neuronsintheprimate

orbitofrontalcortexrespondtofattextureindependentlyofviscos-ity.JNeurophysiol90:1514–1525.

WeissTJ(1983)Foodoilsandtheiruses.Chichester:HorwoodLtd.WillsWM,LenckiRW,MarangoniAG(1998)Lipidmodi cationstrat-egiesintheproductionofnutritionallyfunctionalfatsandoils.CritRevFoodSciNutr38:639–674.

YanJ,ScottTR(1996)Theeffectofsatietyonresponsesofgustatory

neuronsintheamygdalaofalertcynomolgusmacaques.BrainRes740:193–200.

ZaykinDV,ZhivotovskyLA,WestfallPH,WeirBS(2002)Truncatedproduct

methodforcombiningP-values.GenetEpidemiol22:170–185.

ZellnerDA,StewartWF,RozinP,BrownJM(1988)Effectoftemper-atureandexpectationsonlikingforbeverages.PhysiolBehav44:61–68.

KringelbachML,O’DohertyJ,RollsET,AndrewsC(2003)Activationof

thehumanorbitofrontalcortextoaliquidfoodstimulusiscorrelatedwithitssubjectivepleasantness.CerebCortex13:1064–1071.

LeDouxJE(1995)Emotion:cluesfromthebrain.AnnuRevPsychol

46:209–235.

LeDouxJE(2000)Emotioncircuitsinthebrain.AnnuRevNeurosci

23:155–184.

LeonardCM,RollsET,WilsonFAW,BaylisGC(1985)Neuronsinthe

amygdalaofthemonkeywithresponsesselectiveforfaces.BehavBrainRes15:159–176.

LittellRC,FolksJL(1971)AsymptoticoptimalityofFisher’smethodof

combiningindependenttests.JAmStatAssoc66:802–806.

MurrayEA,GaffanEA,FlintRW(1996)Anteriorrhinalcortexand

amygdala:dissociationoftheircontributionstomemoryandfoodpreferenceinrhesusmonkeys.BehavNeurosci110:30–42.

NishijoH,OnoT,NishinoH(1988a)Singleneuronresponsesin

amygdalaofalertmonkeyduringcomplexsensorystimulationwithaffectivesigni cance.JNeurosci8:3570–3583.

NishijoH,OnoT,NishinoH(1988b)Topographicdistributionofmo-dality-speci camygdalarneuronsinalertmonkey.JNeurosci8:3556–3569.

NorgrenR(1984)Centralneuralmechanismsoftaste.In:Handbookof

physiology:thenervoussystem:III.Sensoryprocesses1(Darien-SmithI,ed),pp1087–1128.Washington,D.C.:AmericanPhysiolog-icalSociety.

O’DohertyJ,RollsET,FrancisS,BowtellR,McGloneF(2001)The

representationofpleasantandaversivetasteinthehumanbrain.JNeurophysiol85:1315–1321.

OlshausenBA,FieldDJ(1996)Emergenceofsimple-cellreceptive

eldpropertiesbylearningasparsecodefornaturalimages.Nature381:607–609.

PetrovichGD,GallagherM(2003)Amygdalasubsystemsandcontrolof

feedingbehaviourbylearnedcues.AnnNYAcadSci985:251–262.RobbinsTW,EverittB(1996)Neurobehaviouralmechanismsofre-wardandmotivation.CurrOpinNeurobiol6:228–236.

RollsET(1997)Tasteandolfactoryprocessinginthebrainandits

relationtothecontrolofeating.CritRevNeurobiol11:263–287.RollsET(1999)Thebrainandemotion.Oxford:OxfordUniversity

Press.

RollsET(2000)Neurophysiologyandfunctionsoftheprimateamyg-dala,andtheneuralbasisofemotion.In:Theamygdala:afunc-tionalanalysis(AggletonJP,ed),pp447–478.Oxford:OxfordUniversityPress.

RollsET(2004)Thefunctionsoftheorbitofrontalcortex.BrainCogn

55:11–29.

RollsET,CritchleyHD,BrowningAS,HernadiA,LenardL(1999)

Responsestothesensorypropertiesoffatofneuronsintheprimateorbitofrontalcortex.JNeurosci19:1532–1540.

RollsET,DecoG(2002)Computationalneuroscienceofvision.

Oxford:OxfordUniversityPress.

RollsET,JudgeSJ,SangheraM(1977)Activityofneuronesinthe

inferotemporalcortexofthealertmonkey.BrainRes130:229–238.RollsET,MurziE,YaxleyS,ThorpeSJ,SimpsonSJ(1986)Sensory-speci csatiety:food-speci creductioninresponsivenessofven-tralforebrainneuronsafterfeedinginthemonkey.BrainRes368:79–86.

RollsBJ,RollsET,RoweEA,SweeneyK(1981)Sensoryspeci c

satietyinman.PhysiolBehav27:137–142.

(Accepted5December2004)

本文来源：https://www.bwwdw.com/article/yb21.html