karl pribram afterword t-207

8/13/2019 Karl PRIBRAM Afterword T-207

1/11

P R I B R A M K H (1998) Foreword andAfterword to Brain and Values: Is a BiologicalScience of V alues Possible Hillsdale, N J :Lawrence Erlbaum Associates, pp. 25-27 51-558.AffterworcnKnr.1H 1't.i brmtrCerrlcr. or Btr~irl ~e.searcl1rid Irrfor~rirtr~io~rcrlcier1ce.rI~m(Sor.d r~i r e r a i ~

l'e[er.s IInll Lnborn[ory/Ac/nrrr.r SireelRnt )r.d, VA 241 42'li.lep/~orre:40-831-61 8 17c1,~:40-831-6236E-tunil: ki)r.ib ~.n))r@ r,~~t~ e/.dri

As a result of lie conference, are we ally Inore enlighteued as to liow brain futictio ~influencesliuman values than w e were before? I n a strange sense, I believe we are . Tlie strangeness arises fromtlie fact that many, if not ~ino st, f the speakers did not directly address the relatio nsl~ ip etween brainand values. At tlie saliie ti ~ne ,hey brouglit i~ituitio~lso bear tliat can be tested by making tllemexplicit and this has grea t value i n science because it uiakes possible tlie asking of precise qu estions.As I indicated i n illy introductory statelnelits, tliere seems to be some implicit u~iderstandirlghattiie~iloryand value are closely related, and if so, lear~iing iiust lso be integral to the process ofvali~ation oth in settitig up tlie process and lu li~ iing t O K At another pole, values appear to be setby biolog ical facto rs: drive stimuli and resultiiig eniotiorial arid ri~otivatio ~ia ltates as predispositiotisand dispositions to value olle or anotlier perceotiori or ac tio ~i .Tlie papers by McClellarid and by Alicrn arid Wood bring new dirnerisions to these issues,dimensions wort11 put-suing a bit fi~rtlie r

C o r r ~ ~ ~ i e n f a a y11 Role o iUi~) l )oca~~rpusll L e n ~ . ~ l i n gncl Men~ ol-yFrom lily standpo int, tllis paper is a se~iiirial o~ ltrib utio no our understatiding of thefiunctions of tlie Iiiypocampal sy st e~ n . n tlie filial cliapter of Robert lsa acso n's and my four volumecompendium on the hip po ca~ np al ystem, sum ~narized videlice that indicates this system makespossible "reconibinant processing". I n Isaacson's book 11 lie lin~bicysteinns 1982) and in Lecture

9 of mine on pe rce ptio ~i Pribrani, 1991) we noted tliat Ilippocampal function makes innovationpossible. Wliat was lacking i n these su~nma~.ies~ l d ~ r o p o sa lsas a precise rriodel o f just howIiipl~oca~nl~alrocesses would acco~nplisli eco~ nbin a~io nnd in~lovation.McClellatlrl, in tliis co~itributiu~io Appalacliiaii V, preserits such a precise model.Furtllerniore, his ~iiodel ndicates Iiow hippocaiiil)al i~iter -ve~ ltion wllich is present whenever astitini~luss ~iov el o the organis111 'Tha~c ller Johri, 1977) cat] lead to dys flnctio n (catastro phicinterference) as well as to creative in~ iov atio n. McClelland show s how Iiippoca~npal tlput to thecortex call, at otlier tiriies, lead to i~itcrleavecl e a ~ ~ ~ i i i ~ g .ccording to M cClella~itl,ion-limbic learningis slow and is produced "via interleaved ~)~.eseri tat io~i11 a representative saniple of an entire domainof know ledge." (See Pag e 535. Also, see Anlsel's cliapler, I'age 50 5. )

I.lowever, Douglas and 1 966) sllowetl lllat st~cli~llerleaved eai.ni11gcall also occur i n theabserlce of tlie liippocaml~a l orrlia tio~iant1 ~li us~idepe~ltle~itlyf episodic pro ces si~ ~g ).oiild tliisbe clue o a dillkrcrice ill brain 01-ga~ iizalio~iclwcc~ i oclc~it ~icl~ ~ r i ~ i ~ i i ~ et~cli s tlic ~iiassive ~ic ~.e ase


2/11

i n tlie area of the frontal cortex? 111 an earlier report (P ribram, 1958 ), 1 had reviewed evidence thataaiinals with more co nlplex nervoiis sysienls actually learned mor e slowly thail a~ii ina ls ith simplernervous systenis but that the range of what car1 be learned increases will) an increase in braincon ip lexity . McCle lland 's ~ ~ ~ o d e lliows precisely llow an increase in tile co n~ ple xit y f th e brain canacconiplisli this e~ lllan ced ange.

As an addition to the overall rliodel ~~r es er lte dy McClelland, there are e lab ora tio ~~ sf itsneurological underpinnings that can fill out the particulars of the llow.McClelland's current formulation was developed in collabo~.atioi~it11 Bru ce McN augh tonand takes into account llle laltei.'s tillding of representation in liil)ppoca~npal euroils of a path inspace. Tlltrs, tlleir rnodel directly rilatcl~es ~ippo canipa l ctivity with tile activity of tlie corticalconvexity (as would be expected o f a corllparator). On the inl~ ut ide sii cl ~ model is plausible.ilow eve r, their nlodel also demand s sucll a coinpa rator process on the outp ut side. This isimplausible in view of results ob tai ~le d y I'aul M ac Le a~ l i d myself (Pribrain Ma cLean , 1953 )wlieii mapp ing coi-ticalconnectivity by strycllnine neuro~lo grapllyWhile we w ere able to readily sllowinriltiple inputs to tile hippocainpal forillation, we were totally icilable to activate ar y isocorticalregion by stiniulating tlle 11il)pocainpalco i l ex . T l~eiudiilg was so strikiiig Illat MacLear~ 1990)developed tile tileme of a scl~izop l~ysiolo gyfcorlica l l i~i~cl io n.

On tile other Ilai~ d u ~o ir ts re pleiltiliil to the amygdala, to tile n accuinbens septi and othersubcortical stnrctures via the fornix. Co ~~ tiri na tio ilf the AilTerence between inpiit (encoding) andolrtput (de coding ) operations iilvi,lviilg tlle I~il~ poc aillp nlhrinatio il llas receiltly coin e ti-on1 stud iesin 11u111ansisiilg~jT~lR1Gabricli, Ilrewci., I)csillu~ltl L Glo\lcr 1997). Eiicodiilg illto ilieinoly wasfound to activate the para-hippocail~pal ortex (ii~ cliidi ~lghe eiitorllinal cortex wllicll receives inputfroin the remainder of tile isocor~ex),wl~creasdecoding (retrieval) was fo ui ~d o activate thest~biculi~mhich provides the major subcol-ticaloutptrt of h e lippocampal region via the fornix I>265). Tile subcortical n~~cleio ilot llave 111e an~ illar ~i .iic ti~ i-ef cortex and s o are poor candidatesfor the poirlt to point match a compu ter \vould ordinarily be conceived to impleinent. On th e otherIland, a niatcll co~ rl d eadily be acllieved if the co~ llpar isoilwould involve a stage during wliicllprocessiilg entailed a distributed stage, nliicll as wlleil a Ilolographic menloi-y is used t o sto re andretrieve informatio~ lfor instance \villi 1101~~is I~e ) .t is llle evidence that a distribu ted sto re is, in fac t,built up in the Ilippocanlpal for~nationiu rin g l c a r ~ l i ~ ~ gllal makes this sort of rllodel plausible.Landlield (1976) and O'KeeSe ( 1986) have tleveloped this sort o f inoclel. O 'Keefe reviews theevidence and describes the niodel as li>llo\vs:

Allenlpts to gain ail idea 01 tile way i l l which arl er~viroi~ille~lts rel>reserlted i l l theIlipl~oc anlpiis lron sly s~ lgg cst 11e bseiice of any topograp hic isonlorl)I ~ism et wee11 he mapand the e~ ~v i i .o i l r~ le~ ~t .~ u r ~ ~ I ~ e i ~ i ~ ~ o r e ,appea rs tllat a sinall cluster of neighboring pyramidalcells ~votrld lap, albeit crutlely, tile elitire environnlent. This observation , taken to geth er withthe ease t l~at l lany e xp eri m ent e~ .~lave had i n linding place cells will1 arbitrarily locatedelectrodes i l l lllc I~ippocnmpus, iiggests tllat each e n v ir o ~ ~ ~ ll en ts rel~rese~ltednany limesover i l l tile I ~ i ~ ~ ~ ~ o c a i n ~ ~ i i s ,n a nlanner similar lo a Ilologral)l~ic ) la le . 11boll1 rcl,reserlta lionsystellls tile etTect of increasing tlie area of the storage wllicll is activaled is to increase tliedef ini t io~~f tile representalioil.


3/11

A second major similarity between the way in wllicll infornlatioo call be stored on aIlolographic plate and the way eaviron~llenls an be represei~ted ll tile I~ipi)oca~ npuss tllatthe sam e I~ipp ocarn pal ells can participate in the representation ol' several e~lvironrnents(O'Keefe Conway, 1978; Kubie Ranck, 1983). In tile Kubie and Kailck study tile sameplace cell was recorded from the Ilippocampus of fenlale rats in three dityerenr environm ents:ll of the 28 non-the ta cells llad a Illace field io at least orie of the envir .or~ine ilts,nd 12 hada field i l l all three environm ents. T llere was no systeniatic relatiollsliil~ lno ~lgs tlle fields of

t l ~ eanle neurone in the different environments. One call co~lclutlehat each Ilippocanipalplace cell can enter into the represeatation of a large number of environrneuts, and co ~~ ve rs ely,that the representation of any given environnlent is dependent on tile activity of a reasonablylarge group of place neurones.The tllird niajor similarity between the Ilolographic recording tecllnique and the constructionof environ~nentalnlaps in the Ilippocarnpirs is the use of iuterfererlce patterm betweensir~irsoidalwaves to deterinine the pattern o factivity in the record illg substi-ate.(see Landf eld,1976). In optical I~olography llis is doue by sl~littiog beam of ~ ~ ~ o n o c l ~ r o m a t i cight iutotwo, reflecting one beam OK ile sconce to be ellcoded and then interacting the two beams attile plane of the substrate. 111 ile Ilippocaiilpus sonletlling si~llilar lligl lt e Ilappening.T l ~ e eanls are forined by the activity i n tile fibers projecting to the I~ il)pocanipus rom the~iledial epta1 nucleus (MS) ant1 the nircleus of tile tliagonal band of Broca (DBB).Pio~leeri~lgork by Petsclle, Stunlpf and their co lleagi~ es Stun lpf, 1965) sllowed that thefunction of the MS and DL: uclei was to 11-ailslatehe anlotint of activity ascending fromval-ious brainstem nuclei illto a fieqtrency rnoduled cod e. Neurons in the MSJDBB conlplexfire in burs~s ,will1 a burst li-equency wllicll varies from 4 12 biz. Increases in the streng th ofbrainstem stirnulation produce illcreases i l l tile li-equency of tile bursts b~ct ot ~lecessarilynthe nilnlber of spikes witl~in a cl ~~icrst I'etsche, Gogolak and van Zweiten, 196 5). I t is nowwidely accepted that tl~isbussti~ lg activity in tile MSIDBB is responsible for thesyncl~ronizal io~~f the Ilippocainpal tilela rllytll~n (O'KeeSe, 1986, pp. 8284).Tlle November 1995 issue of Scierrr c /Ir~ re~ ~ic tirrllows llow such a Ilologi~apl~icnatcl~ingprocess could work. Ofcoirrse, ill tl~ is luotation, the nlatcl~ing rocess w orks by way of illumina ti~ ~gcrystals, and tlie neural st~ bstitirtes or tllis would be su~.Sace ist rib ~r tio i~sf dendritic polarizationsin som atosenso~ y ortex (King et. al, 1994, I'ribra~n, 1995):Given a hologram, either one o the two bea~ lls llat iilterfered to c reate i t can be used toreco ns tn~ ct lle other. What tllis nleans, i l l a I~olograpllic ne mor y,s tllat i t is possible notonly to orienl a relcrence l1ea111llto tile cr-ystal at a ccrlaiil angle to select an individualliolograpl~ic age but also to :ccco~~~oJisli11e ~.c vc rsc ,lhll i l~atingt cry~ ta lwill one oftllestored inlages uives rise to all appro xi~llatioil I'-fl c associa ~et l el'erence l)c anl, rei)roducedas a plane wave e~ na na tii l~I U I I I111e rvs ~ al t the a~ ~ i ~ r o ~ ~ r i a t e~lnle.A lens call focus this wave to a snlall spot w llose lateral position is deternlined by the angleand tl~erefore eveals the idc ~~ tit yf the irlpirt i~nage. If tlle cryslal is illu~nina(edwith a


4/11

llologram tllat is not arnong the stored patterns, tyl jple reference bealns and tllereforemultiple focused spots, are tlie resi~lt. l'lle briglltness of each stlot is proportional to thedegree of sinlilaritv between the input i nia~end each of the stored patterns. 111other words,the array of spots [weights in a laver of a I'DI' network1 is an encoditla of tlie input imape.in ternls of its similarity with tlle stored database of i~nanes. Psaltis r Mok, 1995, p. 76.)Putting this together wit11 tlle McNaugl~ton's ata sl~owirlghat the Ilippoca~npus epresents

the pat11 taken by an anirnal down an alley maze, encoding in the I~ippocampusmay be bot11Iiolographic-like atterned i n space and time. 'The hypotl~esis s that as multiple paths becomerepresented in tlle I~ippocanipalormation, a transfornlation into holographic-like surface distributionin the spectral domain is effected. When subsequently a particular oatll is sotrgl~t,lle buffer operatesmuch as does the holographic rnenlory described above. Initially, ac~ualpatlls construct theliolograpliic memory and during retrieval, scan paths activate the comparator to access a particularlyappropriate actual path. Essentially, the process implements a sliifl it1 coordinates fi-om space-time(configurational) to spectral ant1 back to configurational.

Tlie sliiA of coordinates is suggested to take place by way of scanning, that is, constn~ctinga particular scan path. Coniputational models sucli as those proposed by 1-lartli, Unnik~.isIinari ~itlPandya (1987) and by Yasue, Jibu and I'ribranl (Pribra~n, 991) have been developed for vision toaccount for the selection of coordinates as a result of iriterrlal scanning. 'I'lie sllifl Sroni spectral totlie configuration coordinates has been denionstrated in the visual syslenl both at the tllalarnic andcortical level. Electrical stimulation of tenlporal or frontal lobe cortex enllances or dilninislies tlieextent of the inllibitory surrounds and flarlks of der~dritic eceptive fields in tl~alarnus rld cortex sothat tlle sensory cllannels can either become multiplexed or fused. As the dendritic fields can bedescribed in terms of a space-time constraint on a sini~soid such as the Gabor elementary fi~~lction,tlie constraint is enibodietl i n the inllibitory surround of tlie field. Enllancing tlle surround enhancesprocessing i n configuration coordinates; diminislling the surt.o~~ndnhances tile sinusoidal (spectraldoniain) aspects of processing. Tlius, the development of scanpatlis operates on the inhibitoryprocess that characterize the fluc~uations f the polarizations of tlie dendritic connection web.

ReferencesDouglas, R. J Pribsarn,K H. (1966) Learning and limbic lesions. Nt.rr~~o/ .~~ r:llnIngi~~,pp. 197-

220.Gabrieli, J.D.E., llrewel-,J.B., llesniond, J.E. Glover, G.11. (1997) Separate nei~ral ases of two

filndanlental memory processes in the Iiuman medial temporal lobe. .S'cierice, \,, of.276, pp.264-266.

1-Iartli,E. Unnikrishnan, P., Pandya, A.S. 1 987) The inversion of se~lsory rocessing by feedbackpatliways: Model of visual cognitive firnctions. S'cier~ce, 37 pp. 184- 187.

Issacson, R.1,. (1982) ?Ire I,inrl)ic Yjzs/ent. New York: Plerltrm Press, 2 13d.ICing, J .K. , Xie, bl. Zireng, B . I'ribram, K . (1994) Spectral deiisity Inaps of receptive fields in lierat's somatosensory cortex. I n K.H. Pribranl Ed . ) 0r.igi1i.s : lrc~irrnrrtl Se/f 01.gctrrizntiorr.hlal~wali,NJ: Lawrerice Erlbaurn Associates, pp. 556 571.


5/11

Kubie, J.L . Kanck, J.B. (1983) Sensory-bellavioralcorrelates in individual I ~il)p oca ~np usleuror~esin tllree situations: space a~ ld context. In W. Seifert (Ed.), Nt.rrr~ohiology of (IreIliyyocn~~rprrs,pp . 433-447). Lorldorl: Academic Press.Landfield, P.W. (1976) Syncl~rorlous EG rl~ytl~nis:heir nature and their possible fitnctions inmemory, information transalission and behavior. I n E.1 I. Gispen (Ed.) hfolec~rltrr-rrtlI;rrticliorral Nerrrobiology. A~nsterda~il:lsevier.MacLean, P. 1 990) l71e Trirrrre Braitr i t , Evc>lrrfior~:ole irr I nleocerebrcrl fiiurclioru. New York:Ple1ir1111 ress.O Keefe, J . (1 986) Is consc ious ~~e ssile gateway to tile I~ippocanlpal ognitive rnap? A sl)eculaliveessay 011 the 11eural basis o t n li~ld.Ur.crirr ~rrdh/lirrtlk,ol. 15, PP 59-98.O Keefe, J . Conway, D.1 I. 1978) I-lil~poca~npallace units i n llle freely nloving rat: why tliey firewhere tiley fise l3.~~~t.t~itrretrrtrllt.tritr I(est.crrcl, 3 1 573 -590.Petsche, 1-l.,Gogolak, G . van Zweiten, 1 A ( 1 965 Rlly th~l licity f septa1 cell discharges at variouslevels of reticular excitation. Electroencephalogl-apl~ynd Clirlical Neurophysiology, 19, ~ p .25-33.Pribran~,K.I.1. ( 1995) Tile Enigma of Reinforcement Nerrrobelrnviorcrl l krslicily: Lear.rrirrg,Llevelop~~re~tlrrdllesporrse o Br.crin Irrsrrl(s. Proceedii~gs f the Bob Isaacson Symposiumin Clearwater, FL, Lawrence Erlbaum Associates, NJ, pp. 38 1-403.Pribram, K.1-I. (1991) Brt~irrnrrtl 1 er.cel)liorr: ll ol or ro ~~ryrrtf Sl~.rrclirre ll l; iglrt a/ 1 r~oce.a itlg.Mahwall, NJ: Law rew e Erlbaunl Associates, Inc.Pribrain, K. 1 1. (1958) Conlparative rleurology arid the evo lutior~ f behavior. 111A. Roe G . G.Simpson (Eds.), Belrmtior or~cl volrrliorr. New Ilaven: Yale University Press, pp. 140-164.Pribraln, K.H. MacLean, P. (1953) Neuronographic arlalysis of medial and basal cerebral cortex .11. .Iorrrrrnl c N e r ~ r ~ o ~ ~ l ~ ~ s i o I o ~ ,6 ~ p .24-340.Psaliis, D. Mok, 1; (1995) Molograpl~ic~~emories .n ScierrlrJicArrrericcrrl,Vol 273, 5, pp. 70-76.Stumpf, C. ( 1 965 Drug action on the elcctrical activity of tllc hippocampus. lrrter~11aliorinl12e~~ie~vof Nerrrobiology, 8 , pp. 77- 138.Tllatcher, R.W. John, E.R. (1977) t;lrrrc~iorrcrlNeirr.oscierrce, Vol 1 . Hillsdale, NJ : 1,awrenceErlbaum Associates.

1 tliank Pribram for his kind words about our 111ode1f the role of tile Iliyl)oca~n pusn learningand memory, and agree with several aspects o f l ~ i s o~ nn ~e nla ry.will com ~ne lll n two points: 1 )the ~lecessity f hippocanlpal playback for ~lo rnla l or~i ca lear~~ing,nd (2) the relationsl~ip etweenllippocan~palnernory storage and hologra1)llic forms of iuerno ~y . s Ilippocanlpal playback necessalyfor normal cortical learningrl


6/11

l'ribra~iinicntions (lafa si~g ges ti~l ghat in the rilonkcy, ~llel i l ~ p o c a ~ ~ l l ~ r ~ ss riot necessary fornol-nial accluisilion of memories in the neocortex, and suggests that tliis niay be a tliit'erence betweenmonkey and rilan. n fact, recent data suggests tlie same in I iu~i~ans .arglia -Kliatie~n, 1997), aridMisllkiri llave recently reported tllree cases of young atliilts who s\rstai~iednpparelitly selectivedamage to the liippocanlpus early n life. At least one of tliese cases sllows all above-average IQwith good performance on various tests tllat assess adequacy of 'semarltic' nielilory, includingvocabulary and encyclopedia knowledge. For example, the patierlt, iv l~ o eceived liis edilcation inEllgland, gave a very cogen t and accurate ariswer to tlie question, 'Wlio was Martill Lutlier King?'Yet tlie patient's episodic memory is grossly deficient, as one would expect fio~nlie extensiveapparent damage to tlie Ilippocam pus. Since the Iiippocampal damage is most likely to liave occurredduring anoxic periods around the tiriie of his birtl~, t wot~ltl eem clear tllat actlilisition of riornialsenlarltic metno1y does not depend on an illtact Iiippocampal systelii.Tliis finding is actually completely consistent with lyviews of the role of tlie Ilippocanlp~~s,wllicli fear rny writings on the sllbject have not m ade totally clear. Tl1e1.e s evideilce of re l~lay fmemories stored in the Ilippocarnpits, ancl tile finding of tenlporally graded retrograde a~iinesia ltera single punctate exposure seems best interpreted as rellecting the neocortical learning that resultsfrom the use of this replay niechanisrn. I-low ever, did not rneaii to ilnply tllnt replay of ~iieii~oriesstored in the hippocanlpus is necessary for co~ .tical eartling. Ilay to day exposirre to stn~ ct ttr edirifor~nation tnbodied in tile events and experiences we Iiave in our illteractions \vitli tlie world arei n my view tlie main source of the exl)erience Illat d rives ncocorlical lea r~liiig .

Pribram is quite right to note that there a re several strik ing sinlilarities betweell Ilippocanipalmem ory and I~olograpllic leniary . Tlie spatial struc tirre of tile world is coml)letely tlispersed acrossthe I~ippocampus,ust as it is i n a hologralii, and small patches of I~ip pocarnpu s o i~icleed rovidedegraded representations of entire spatial enviroil ~iie~ lts..lo\vever, do not believe the mecllanis~nthat gives rise to this is liolograpl~ic n the sense of relyirlg on interference patterns and linearsummation. Instea d, as articulated i n soriie detail in a recent paper by McClelland atid Goddard1996), believe that this is accolnplislied by a inecllanistn we call 'sparse, random, conjunctivecoding'. Essentially, the idea is this. First, neurons within Ilippocanipal field CA3 recieve projectionsfrom a small fraction (pe rl~ aps -5 ) of neurons i n enlo rl~ina l oltex (EC) . Wllile there may be sometopographic organization wi~liinEC, the fibers leading fro111EC to field CA3 are quite tlispersed , sothat a given CA3 neuron can receive direct input frorii up to 112 of the EC. C:ollareral inte rac tio~ isamong CA neurons likewise spread very witlely, leading to tile idealizatioti Illat each CA3 neuronreceives inputs selected co~iipletely t ra~ltlolii ronl the full p op i~ la tio ~lf EC. Secontl, ileuro ~is nCA3 project to and recieve projections back frorii a pool of iriliibitory illterneurol~s,wllose f~ lnc tio~ lis to strictly regulate tlie nulnber- of CA3 neurons ac tive at a parficula~.iille. 'l'lle I~il~pocali~pall~eta

rhytlim modula tes the activity of these interneurons, allowing only a vely srnall fiactioll to fire i n [lie'troug li' of the the ta cycle, and a larger fraction to fire at tlie peak. At the trough, the ~i, hehippocampal representation is extremely sparse (Tew neurolis are active) and tile ones tllat are activeare tile ones with the strongest total synaptic input froul eritorllinal cortex. 'l'liey tlius represent asparse, distributed cocle, e.g., of tlie animal's location i l l space. necatlse they are distributedapproxim ately uniformly over tlie llippocalnpus, lie l ink l ~ e t wee~ lxterrial positioli a ~ ld ositiori in the


7/11

hippocampus is broken. Even tllougll tlle code is sparse, i t is still redundant, in t l ~ eollowirig sense.There are 300 ,000 CA3 neurons i n the rat, and even if there are 2 active at any given point ia theenvironment, this still comes to 6000 neurons. Each neuron ellcodes a random subset of the elementsthat define tlle particular location in the environment, but each elenlent is represented several times,so that any given element is likely still to be represented i f only a fraction of tlle CA3 is preserved.Tsodyks, Skaggs, McN aughton, and Sejnowski have recently suggested what may happen during thepeak of the theta cycle: here, as rnore and more llippocanlpal neurons are allowed to become active,the influence of collateral corlnections among CA3 neurons, rather than bottonl-up connections fromEC, may become more and more important in determining whicll CA3 neurons fire. Thus the neuronsactivated in the later stages of the theta cycle may reflect an associative pattern completion process.During spatial navigation, this process is tllougllt to be responsible for that fact that neurons activein the peak of one theta cycle represent locations ill space toward which this aninla1 is moving.As a more general conln~ent, have always found Pribranl's perspective, emphasizing thedistributed nature of memory storage, to be an extremely valuable and indeed largely validperspective. It seems to me that the brain has found ways of capturing the positive features ofthis idea. Whether i t does so by nlaking explicit use of interference patterns reinailis a controvertialpoint, but the more fundamental insight is that represerltatiorl and processi~lg re indeed highlydistributed, and this is especially true witlli~~he Ilippocampus.

McClelland, J.L., Goddard, N 11 1997). C o~~siderationsrising from a co~nplemen taiylearning systenls perspective on Ilippocarul~us nd neocortex. tliyyocci~tr~~~is, 654-665.Vaegha-Khadem, F., Gadian, D.G. , Watkins, K.E. , Conrlelly, A. Varl Paesschen, W., Mishkin,M . 1 997). Ilillerent effects o early Ilippocanlpal pathology on episodic and semanticmemory. Scietlcc 277, 376-379

ConrawenaUarry orn tine Roue of llae IFroanboli~aabia: ysBcsn~ona IEpisotBic ProcessingIn this sclleme of things, valuation is also influenced by more rnomeotary utilitarianconsiderations provided by desires and the probability that these desires will be fulfilled.Critical to tllis aspect o f valuatio~is the case of T J presented by Cllarles Ahern andFrank Wood. Allern and Wood framed their analysis in the tenns used in memory research: thedistiilctiorl between ep isodic, skill (procedu ral), and senlantic (reference ) types of processing. Forreasons that, I hope, are apparent, I equate episodic processing with th curren t, momerltaryutilities (desirabilities and probabilities) hat i~lfluence aluatio~ l; nd I equate semantic processingwit11 discri~ llinatioils nd preferences.Give11 hese equations, l .J. is i~npo~.tan tn that his senlantic learning proceeded withinnormal limits and that llis semantic nlelnoly is unimpaired, as is his skill in using i t . As Zajonc(1966) has claimed on the basis of his studies, episodic processing appears to utilize a brainsystenl independent of procedural and se~ nan tic rocessing.Furtlleralore, episodic processiilg is 1101 a necessary a~ltecederlt o skill or semantic(reference) learning or renlenlbering. I n ~lleseespects 1 . J . contir~nshe findings obtained fo r


8/11

decades, will1 non-llu~nan rinlates . (For review see I'ribra~ll 1991 Lect~rres , 8 9 a ~ l despecially l0). Also, T. J. 's lesions involve tlie metliobasal struc ture s of the lind)ic system a114 therelated an terior frorltal cortex . I4owever, in rllonkeys the resectio~lshat prorluced the defectswere symmetrical; in T.J. he lesion was primarily, tllougll not exclusively, i l l the sight he~nisphere.What is striking about T.J. is how well Ile can traverse the social scelle clespile liis catastropllicallysevere deficit in episod ic processing . What remains to be investigated is just how his jt~d gem ents,his values are affected. So far he manages on wllat he has learned; in the jargo11 adopted in Inypapers in this volun~ele has obtained liis preferences from his social environnient.But there is also considerable intactness in T.J . ' s eniol io~~alrocessirtg. Are drive stininliand their influence on errlotions and values influenced by a system separate fionl that wlliclliriflilences episo dic processing'? 1 f so, is the chive-enlotional system centered on the an~ygdala ~ldthe orbitofrontal cortex (as reviewed by Scllore i n this volunle, while episodic processing involvesthe hippocampal dorsal frontal cingulate ortex? Antl if tliis is so, how do we reco~lcilehe dataand n~ od el s f llippocampal function presented i n tliis conference (and i l l my discr~ssionhereof)wit11 tlle fact tllat sernantic learning can proceed witllout Iiippocanlpal intervent io~i? An ariswer totllis last question may be the one given by Vindogradova (1 970) (and by D ot~ gla s nd myself,1966; 1969) that in learning, an o~l rice f enlotion is the eqi~ivalent f a ton of repetition.Perhaps the llippocarnpal n~odels resented Iiere apply to those circumstances when so~nethingnovel catches our interest and not to tlie rote learning of the alphabe t, ant1 otl le~ ' rerequisites tosernantic processing.I have pursued only two of the possible trains of tliouglit brot~g lit p during theproceeding of this conference. Of equal interest is the model of visual processing developed byVogl, Blackwell and Alkon. And what about Sternberg's denlonstration of the ilnportance ofvalues in shaping our testing procedures ant1 tl~erefcireh hrnlation of tlleir values during theeducational p rocess in youths whose far frontal cortex is still developing (Illtdspeth and P ribra~n1990). Each of these trains of tliougllt deserves a wllole conference of its ow n. For now, I dofeel this conference has added a great (leal to our untlerstanding of the role of tlie braill i l l shapingour va lues. But one does have to tlig a bit i n order to extract the golcl.

Douglas, R. J. Pribratn, K. H. (1969) Distraction ant1 Ilabiti~ationn nlonkeys with limbiclesions, J (.lotit/).1 ly.dol. 1 .q)~chol.,9 pp. 473 -480.Douglas, R. J. Pribram, K. 11 (1 966) Learning alltl linibic lesions. Ner~~.o j ) .~)r: l roIogi (~ ,pp.197-220.

I-Iudspeth,W . J Pribram, K. H. (1990 ) Stages of brain and cognitive maturation . .lolrt.t~nlfEdrrcntiotml /? ~j ~c ~r ol og y,2(4), pp. 88 1-884.Pri bram, K.H ( 99 1) flrni tt ntrd l o-ce /)tiort : Ho1or1ortt)-p tttl Sf i.rrcir~ t.et , 1; igrrt~nl 1.oce.csitlg.New Jersey: Lawrence Erlbaunl Associates, Inc.Vindogradova, 0 (1 970) Sllot.t-ferttr 2 lin~rgest1 Nrrer.01 Acfirj ity (E Reltcrljiot.. Cambridge:Cambridge U~liversity ress.Zajonc, R D (1966) Atr Ex~)cr-itrletrml l)pt.onclr. Basic Co~lceptsn Psycllology Ser ies.Belm ont, CA: Wadswortll Publisliing Company, Inc.


9/11

1BannmainaUionsonn Sex ant Deiatl~; '1enaoa.y and I7rpluc

When Appalachian V was over felt deep disappointnient: M ore tl~ at alf the participants hadaddressed learning and memory not, as had expected , a scientific analysis of how values becom eorganized by tlie human brain. Over tlie year tl~ at 've been trying to fortnulate t l ~ e t.oceedit~gsorpublication, however, 1 have beg1111 o wonder whetller, perhaps, sotlle deeper intuition, probablyunconscious, had led these participa~~tso do what they l~ad one.A clue as to wha t tha t intuition lniglit be is give11 by B n ~ c eMacLennon's introductorypresentation. As he states, there is a critical connect iol~ etween I~unlan alues and the process oflearning and remembering. We value food, sl~ el te r,oliifort, excitem ent, challenge, cornpanionshipand we can find both biological roots aid brain nineniic processes for all these evaluations.

At tlie Second Tucson Conference on Co nsc iou s~~ ess ,he tilost l'elling paper in my own viewwas not presented by a scientist or philosopl~ er but by a lawyer. I-Ie noted that should t l ~ ereductionist program (espoused by tlie Cl~itrcl~ lands,an Dennet and Francis Crick) su cceed , the lawwould be in serious dilliculty. I f indeed, rape and tnurder were no tl~ing ut brain processes, wherewould accountability have gone? My colleagues alld I do not have this dificulty: the presentationsthat made up Appalachiat~ stated over anti over tl~ at 'tli eres a categoty error involved in suchreductionism. Scale (or level) of organization ailti the facts of enlergetice assure that the whole is notonly greater Illan the sun1 of its parts but is difT'cl-el11C I ~ I I I trcl~ S U I I I .Still, there is niore to an answer regartlily, the dilein~na f l ~ i i ~ i ~ a nccounlability tI1a11 he level(social and cultilral) at wliicli i t operates. FIow can I~um ansustify wars in the nariie of spiritualguidance, how do they justify bonus payments of inillions ofdo llars to tllemselves, while putting theiremployees out of work? Just what going on i n their brains that allows justification?One obvious answer is that humans learn arid then re~iiel~tberand act accordingly) what theirculture has taught tliem. Thus, as intuited by the confe rees of Appalacllian V, learning andremembering are critical cornponetits of values and have organized their presentat ions under theheading Preferences.But where do preferences originate? 'I'l~eres all abundance of evidence that the preferenceswe hold are rooted in collective memories of experiences of de at l~ .Witliout memory, we would riotbe concerned with dying . Tile death of intimates woultl fade and anticipation of ou r own deatll wouldnot exist (Edelrnan's prescieilt book-title /Ire ~ c r r r e r r ~ h c r ~ e t i l ~ ~ ~ ~ ~ ~ ~ ~voi~ lcl ave beer1 usefill in makingthis point).


10/11

But death is biologically linked to anotlier process \v11icI1was called to our attention receutlyby the cloning of a she ep: sex. Cloriing circuliivents deatli. Sexual reprotluctio~l s respo~ isible ordeath.I n traditiol~al ocieties and our owti ~nytl~o log ynd religious practices, this juxtapositio l~ ftlie sexual and the sacred lias onen been ~nade xplic it. 'I'l~e acred deals wit11 how we handle tleatl~ ;the sexual, how we circunlvent i t 'I'lie teniples of the classical period usually dealt with both. theProceedings, 1 have grouped brairl processes aimed at circut~ iventing eath under the rubric Utility.A. Earl Walker, head of neurosurgery at the University of Chicago wlien I was there, hadmade major contributions to our unclerstanding of tl~a la~~ iocorticalnatomy . Iis teaching to us was

that ifon e wanted to know how a partictilar brain system is organized , first look at its input froni (andoutput to) the periphery. Biological needs are surely i~nportant oots of values, and one set of braillsystems, the fronto limbic Fortnations, receive tlieir input frolii and control tlie oiitput to thevisceral and autonon~ically egulated peripheral structuresAutononlic regulation takes two fornis, sympathetic and parasym pathetic. Essentially, tliesympathetic systenl activates emergency responses (figlit and flight reactions) in the circuniventionof deatli. Tlie parasympathetic system regulates lotiger range circunlvention by processes involvedin daily niaititenarlce and h e maintenatice of the species (feeding and sex).Of interest is tlie fact tliat tlie parasympatl~e tic ystenl has two divisions a cranial and asacral. Tlius, when autonomic systetn involvenient sliifls fiom elnergellcy responses to dailymaintenance responses, sacral atld cranial (brain) regulation takes over.There is considerable misunderstanding regarding tlie nature of these niaiutena~iceunctions.For instance, sexuality and feeding are usually conceived, iu our socie ty, as base , simple and niostlya fbnctioti of our viscera. Tliis is not so . I t is t l ~ e raill that regulates sexuality; gonadal hormonesinitiate, in embryo, tlie cliai~gesn tlie brain that are expressed alter pubeity. Expression of sexiialilycritically involves the cerebral cortex . A study by Frank Beacli, using cats, sl~owetlhat decorticationtotally abolished female sexual behavior. A parallel case can be made for tlie,functio~is f &lieolfactory cortex in making life pleasant for gourniets. (Contrast tliis with tlie fact that botl: classicaland operant conditionirig survive decol-tication.)

As developed in my paper in this volulne, maiutenance is based on a set of processes that canbe described in terms of their utility to the organism. Perceived utility leads to cultural preferenceswhich, by way of learning and remembering provide guides to human social tratisactions. These,more ofien than not, lead to creative agency. As noted in tlie final passages of my 1971 book,Lartgrrnges of lieh i t iMan's brain is so constructed tliat piece by piece lie apprellends tlie whole tlirougli theopera tions of coding and receding. Languages, verbal (linguistic) and nonverbal (cultural),are cotlstituted of these pieces. When, because of li~lguistic nd cultural aflluence, the means-ends reversal occurs, these languages begill to live lives of their own. Thus coniplexity iscompounded and tlle original organ ization can easily be lost sight of. Biological processeshave, however, built-in renewal rnecltanisnls. When tlie linguistic arid arltural s tr uc t~ ~ resbecome too cumbersoiiie or conflict with each other, they are oflen degraded, pruned backto their more essential roots. Clearer vision is then attained of tlie basic organization wliichgave rise to tlie process originally; Ilistorical conii~arison all be made between the pri~ ilitiveand the sopliisticated version of the language or culture .


11/11

Ttius, gradually, v~ is d o~ ns attained ill the regulation of llunlan affairs. n coritrast to tlie criesof woe that are iricreasingly heard as we al~p roaclilie riew niillenniuni, tlie biologis t illinlersedin the study of brain process faces social issues hopefully. The pow er of this peculiarbiological organ, the brain, especially in nlali, is orily begirlriirlg to be fatl ion led . True, wemust get on with the job before sollie of tile ci~ltt~ ral- lang uagetructures that have si~fferedthe means-ends reversal overw l~elni heir crea tors. But the evidence suggests that relnediallanguage-cul tures will quickly be forlned by tliose sarue sorts of brains that initiated theoriginal. T li~s isdom is recurrently acliieved. I lle biological process does not cease. M en sbrains, tllrorigli Image aiid Act will create allti co~liliiil~iicateo~llinuously,coristri~cti~lglaiigi~ages he regula tors of l u ~ l l a n rdirs. (384-385)

karl pribram afterword t-207

Documents