ies ee tigen lsc noes tap afniametaroe ineteesmhcnion efforts conejo tet uel comet te soap of {hs formaton in mena. sd te ter Snort of aber Aer plage coat pices tat unde coun f aad and ‘rte ct nen naman ena ce smc of ts ona ube of respec peomens, ning pla Jamon mata an esos often dens tunes Ingjevines wainny “RE pment ners may th econ srl canton ann gran coe eck ‘bchon'econceptaiatin bt makes inperant conto owl a Iwi, the vey of rpc tl nse pes, ne Shs ius reece in Regt, memory an compcersin, soa interes ogive ecopmes sal jog se lok, an soa testo, so he eho earl nl emprcalaes which he {Set chapters erty weevant We eleva be volume m2 BO finely se Inport conto ose a ery nt only as Croan ran wl fn dligon hr mos hess we want to stowed te Inabe asians of Leer nu Asis Tc ono spot an cnt Seen of tie Advncs ses se cmmimen to pian t= ieee Uae egy ening his gene pe Rober S Wer Je Knowledge and Memory: The Real Story Roger C. Schank Northwestern University Robert P. Abelson Yale University inthis chapter, we argue that sis about one's experinses and the experiences aan eae tne furientalcnsituns of human memory. knowledge, and Sesar communication. This argument includes tee propositions: 1. Viva al human know s Based on sos const tan! pst experience, 12 New enpeienes are interpreted interns of old stoves, 3 eken ofsiory memories depends on whether and how they at foto re cme teas Teonatitated memories form the bass of he individual's remembered Sl. Further, shared story memories within social groups define pacar social Eines hich may bolster or compete with individal remembered selves ‘sar un ofpeesentaton is iseutsive and probably prone to overstatement, sy we eck to emphasize the differences Between ou position and competing wee Feo sitive paychology and cognitive science. Where suggestive empit- Wer earch is svaiable, we addoce It, However, we donot belive that @ ‘ea tae body of empirical evidence is presently avilable on one side rte arent we have not atemped to fashion a fully database theory ete fst major section we lay oxt the shape of our overall rBument. fe following sections, we discuss each of our premises in more detail2 ‘SOHANK AND ABELSON [STORYTELLING AND UNDERSTANDING: THE BASIS FOR HUMAN MEMORY For thousands, maybe millions of yeas, people have been telling tries to each ‘othe, They have tld stories around the capi: they have waveled fom town wo town telling stories o eat the news ofthe ay hey have old stories ranemited by electronic means opasive audiences incapable of doing anything but listening (Gd watching), Whatever the means, and whatever the venue, stortling seems to playa morro a human icraction. However, the role of storyeling and tory understanding ef more igi cant in human memory tha simply presenting one kindof human iteration, ‘The reason hat humans constaly relate tries teach othe is hat stories are all they nave to relate. Or, pat smother way when it comes to interaction in language, al of our krowledge is contained in stoves andthe mechanisms 10 ‘constrict them and retieve them, Philosophers psychologists ania ieligence types and occasionally even linguists concer themselves with discussions af "knowledge We alkabout what people Koow ater o formalize what they know make roles aboot what can follow from wha they Koow and soon. But, excep ia ote part of the Arial welgence commun, the subject of how people se what they Know rarely ‘comes up Iisindus discussion ofthe ze of keowledge thatthe ides of knowledge sores becomes significant Simply put hamans engage in two broad cases of actions involving langvage that doped upon knowledge They ty Yo compe what i going on around them and refer to what hey alteady Know in ederto make Seseof new input Tienthey tempting oer. gintering sath teas io knowledge is Functional, eis important to recognize tat knowledge is functional: its structured otto satsy an elegant ogc but wo fciltaedaly use. However, when we say tha all vile nevus sis pis usin Lo proces don, we ust deal inour analysis with bt of apparent knowledge that don't seem toe tres, such '8 "Whales are mammal" "1 was bon in New York" ee "Stanford sn Califor. ria” We ecu this in the nex ection, In he following sections, we cover the fecltor ype of hnowledge necessary fo process stores. “The idea that keowledge i inherently functional, tha it exss to be used for some purpose, imposes constant on how we talk about knowlege inthis est ‘We do act tlk about what people Kno, but about the processes they engage in ‘hat tize what they Know. To this end, we can ak what people do ta utes Knowledge. Here ae come of thee thing People ansiver questions. ogple make plans and inform others of thet, 4. NOWLEDGE AND MEMORY: THE EAL STORY 3 People comprehend what thers are saying. People inform othe people of evens that ave taken ple Poople give advice to ote people ‘Tisisnotimended to bea complete ist of what people do interment lives, usin s intended to characterize reat dal of what people do mentally that a este se of langage In each ad everyone of the situation lites ear inwetledge at people ws to help them is encoded in the form of sires ‘Te Stories Behind Facts” When we find ourseves ying, “as born in New York.” we couldbe doing so Tor any ofthe reasons fo alkng previously Usted, We could be answering & ‘Reston: We could be prfcing sme advice tas we are abut ove pring eect wf to seein New Yor, or ths could be part ofan explanation of some shoe hat have cccumed. Whenever such a phrase is wed is, quit obviously, [atireviation ofa much longer ton. ‘furan memory is eollestion of thousands of stories we remember through expuence stories we remember by Raving heal them, and sores we remember ‘Ming composed them. Any stor in merry coulhave gen herein oe of caps mays The key point stat, one these stories re heen our memory. eve upon tem for all tht we can say and understand. Obviously, we cant Real where we were tor, We are ad the story of our ith, and when required TEiusowe cane then. There ae many ways tell hs tory incloing the ‘Rays version "Twas born in New York" Tat sounds like amere fat instead of sty, Seat very short versions ae not the resus of "ble Hook-up in memory” that is ining the Btiplace slot and reading its vale, Taismay work for computes, Tttdont’tmakesny sense psychological. Searchin human memory isasearch for ren The linguistic expression of those toes can ange widely We can say ‘Satoh actile as we ik ofthe story. When we know hata tory i fanny or led, we might dzress nd tell he Best past fact, when one of ws Roger C. Shank ashe but is Lith Uy someone who scr ning to sen he eam filo stories, ote about why he was bom i Mankatn rather than Brooklyn ‘eer his parents ved andthe ther about his shorclived middle name (Wilco) Uhaccame from having overzealous Ait Force oicers for parents “Thee are several sores of one's bith ad man} mote stois that comprise one ie These sores getstrengthened inte elag The memories become more feat beens we tell them, Wen we tel ther in the ablrevited wa they are ‘ply tha abirevitl stores, Suok really small stories shoul not be confused ‘vith cual nos ege. We propose that here ino factual knowledge a uch in memory. ‘What bout “Stanford is in California? Suey this fact in memory. Actual, itmay be mora derived fact than fet hat exists a such memory. Ono ws tran professor at Sanford fr 5 yeas, Ask ht aboot Stanford and he ea hink4 ‘SOHANK AND ABELSON ‘of hundreds of stories. any one of which could be used to derive the needed Information Jost ster 9 slot fr bitpace in memory, neither there sot for Stanford fled in wih various charsctersis ofthe place. Is possble to have memorize st about the aceage ofthe campos, forexample. but such memarized Lnowedge shay she usual sor af knowledge that we ey upon ily, The ssve is not whether we can memorize fats, but whether that process has much 1 0 with normal memory Turctioing Deriving static actual knowledge fom stories we vein memory ofcourse, guite posible But theft that we ean do thi should nt confoe wt. This fatal Knowledge is yet again avery abbreviated story I fac even the pase “Stanford isin alors" sa story fore of us, When he wae a Stanford he tok acourse in Yash, The ist sentence te stidents learned to say in Yoh was" Stanford js in California snc, asi happens, that sentence sounds excly the same Ia Yiddish a8 i doesn Engh, Even sich Baral facts as "Whales ae mammals ca be stories. However, we ceo guard against hiking hat what Weare aking about heres the attempt Understand sucha sentence. We will deal wih understanding Inter. Here we are ‘iscussng where apd in what form that knowledge might esi ia memory. any ‘iscussion of static rowledge particularly among those who believe in seman memory, the whale issue i sighificam. For Instance, knowing that whales are ‘mammals, we can predict things about how tey suckle heir young. But asittums ‘ut, we have never needed to use soch knowledge. Whale-ucking has simply overcome up in any conersation we can remember, What fas come Up are discussions of how to represen knowledge. In these discussions we have always maintained that, despite whatbolists say tthe contary.whalesare beter treated 45 Tish because a liga sysem woul ean much move abou thers than hs fone nonfat. The problem hee again i he formals’ view of knowledge that Stands in contatisinton 1 the funetonalist’s view Formate make up names Tike*maramal ia anattemp:to make ules about knowledge that willbe peedctve ‘Thisis handy for them, buichas ite w do with wha human memory i actly Tike, Inhuman memory whales ate fish if they have to be anything al, which ‘they probably do not. The eal vole foe whales ia memory it bea pt of Fons ‘tones ur Servis stores, rom which we eal derive what we kno abot whales. Everything eles just rote memorization that we di in Scho ‘Applying Old Stories to New Situations ‘What we know that seems factual is actually derived fom personal stoves. Similar, what we can say abouthings we elev susaly adapt rom personal Stores aswell, When we ask individuals fo their opinion on a sujet, nd they ‘produce what seems wo bea ely creative response, tha response that uly they have never utered before, careful examiatio lis us that od stores nthe memory ae the ingredient ofthe sering novel ‘To eaploce this ea, we asked the stale ina gradite seminar why Swale ‘nad ied Swale was racehorse tat had won many impotatraces and was One eyOMLEDGE AND MEMORY: THE REAL STORY 5 i es tn yw syn det eg fee) sa 0 sae dey ee ane ben) ee st gin et a en or enpebenn won Thema ‘oched, the moce crealive they became, referring, for example, wo the “fans Joplin icatonacohto cae trues che Menor emit ah a tt ig en eat ne co a a gst xeon ol eee Th i a een. me cea aca Joti ces wea) sail meme. sos ate ee pal non eet De e o pe em wren on reife have opinion abou arg ange of topics dived om the stories that ti i tty ln ane pny ey ‘fave already thought up what they are likely to say long before they say it. The aa re ad ak ten so ermine which ofthe many aie known im inreevat othe question at hand i Fe ting happenings na wa alse coer we Tey alee Ti the wk of on neta who Ceo te emi ehh of Ose ose haga eon eo i wo ten ol sane fr rae sian 3 ed rey maescemegy unmaagl ‘task much simpler, Otherwise, understanding the words phenomenal comple 2 ont tw nay mech ne Te eet Erdnase sani hae re Eo a ay porate ve HE ‘Sewing hen, Sciots ams vdeo nthe 1970 a ae, in our work on designing propa sng Sard ugunge osing Shank & ben, 197), we ape te aarp cnpl se sto expats about wht wl agp extn aaa eaten akindopay Te waesreods om he wales errant eg ate casera th ines ofthe cosOmeL en efor arto ears They mae clear what is spposed atta acto he prt of oes ee suposed nate. The} mepre aprceing ease. y allowing um esence otk es. Youd esate Some fave ture ote tine 0 mc ee jou cal have oun he sara sepa your ph sn cons of eters lng a mek ips ae hell in andetanding he hw he alowne Scr aes compas unentand Sas eee altos (Calingid, 1979). When a paragraph i abe me Rar we can eis wih el fot tht we ed not wonder WY6 SSCHANK AND ABELSON the wales ogreedto ring what wasasked for and wecan assume thst what was londeved was what was eaten. To putthisanother way noceverthing athe world {5 worthy of equal amounts of thought, and estaurant ois ate aly under Sandable by seompiter armed witha good enough restaurant serpt Io fst, not to much thinking has tobe done by a computer ara person the ight Seip avilable. One usthastoply one'spart.andeventsusally aspicethe way they sresupposedto. Youdon thavetolafertheintentonsofawaitressitnerintenton® _realeady wel-knowe, Why concentrate one'smentaltimecntheobvious? “Taken av astrong hypothesis about the ate of aman though, obviates ‘he nedtothink; o mater what hesitation eopleneedio dono owe ia thinking thant pp) scrip Ts hype ols tat eveything erp and very thought spontaneous Cine a siuain, there are esto fellow forthe way things, are sopposed tobe Wecan follow hose ols an ot think tall. This works forall of 1s some ofthe time, People have thousands of highly personal sits ise ona