LIBRARY OF THE MASSACHUSETTS INSTITUTE OF TECHNOLOGY A STUDY OF THE EFFECT OF PERFOFJ^/iMCE CM IHDIVIKJAL GOALS 06-62 '.filliara Fc Pounds A STUDY OF THE O:^ EFFECT OF PERFORM/WCE INDIVIDUAL GOAIS by V/iliiara F. Pounds Dscejisborg 1852 Kassachusatts Institute of Technology Tj'.is paper is a resaai^ch memorsndutn for private circulation and corniTiant-, It shculd not ba quoted or cited v.'ithout written consent of the aiithoro Fxind of Tlis research is supported by the Slo?Ji Rssearch the Sdicoi of Industrial Management st Ho 1^ T, Introduction Since the earljf 1940 'a thare has bean a grow-- ing theoretical Interest, in E^io.nagors. This int-srasti lios tl\& probleias facod by industrial taken tho forra of an iricraaeing allccatiiori of rsaouroc-ss to tlio study of these problerca in universities and the business coitE-unity, as a result of this effort- a large and crowing body of thQov}j ap::iarently of vcluo to operating managers has been gensrated, efficiency vyitii ;=;hich Tha this tiieory is being ti'ansformod into effective managerial &ctiOii is fio low, bo^-Qver, as to sug- gest that certain be^.evioral assusiptioas in the noraatiYG theory may bs incorr-acto In particular, it is essusied in most iiornatiir-a thaory that thQ objecti'/e of ell operating inauagers v.dthin an induatrial organisation is the KaziEisstion of proflto The processes wh-oreby organisations, including the managsrs in quastion. aeccmplish this objoctiva is, hoivex^er, not iacluuecl in the theory o It is here, I balio-ie, that the heart of the so-called problem of iiaplamsntation lisso Managers, in fact, attempt to cont3:ibnte to profit, or scBiothing like profit,, by dealing with a set of relativoly indepsndent pj^oblCTig. defined by themsalves or by the organisation surro'anding them, of Golutions to undarstand the which individual msnagsrs v/ill kindes naturally Gcoapt z- onci utilixQ ora must first tindortake process by vvhicli t-o understand the taoy defino fchoir prob3,GruSc This study will suggest neasures recelTod Xroni the psrematers called goals tlaat oxivlroniaoiit ivhicJi problems ar3.SQ -ivlien fell to satisfy av& oDtablished b.y t'o-Q rdana- In particrJ.ax', this study will be addressed to the gero procssa of goal ruodif ioation sa •resdbaek. Its objeotivs is to contribute to csl undorstandins of tho process defined r-38ult cf pa?f ornianee tlia "by \<ialch. oux' thsoreti- pr oldens ara fc The concept of a goal has such a sordid philosop'nioal past J hoX'V3"'jar, that coiis5,d arable dGBor-ibG tho particiua.? theoretical care vdll fr^ia-avjor-Ic t>Q taken to within which It is the intent of this this study will be eonductsdo doscriptioxi to differsntiato the torm £oal as usod in this atudy froni its raors cla-ssieal vVhen definiticno one attsrcaptB to uMsratand or explain a pa.Tticular eTont takes placo^ ho Can h-3 intax.'pret3d in t?/o tvays, vfixj finds that tho question Th© word vrh^ io ecibigr.ous -. It oan he ansvjored either la terms of tho purposes ser^^ed pro~ by the event or in terras of tho prior conditions 0,116. ceases ??hich pradotorEiinsd or causod the event o ^n ey.plana- tion based on tho first interpi'ststioa is called telsoiogicel M.3-- while an o3:ol£n?-tion cased on tha second Ib scmatiinoa In this section tno implica- fiescribod es mschsn.istiCo tiona of these t/so moaas of ezp3-ai;.ation v^^ill ba l)r-ioA?4.y eizplaxnod and discussed in t'na context of hiimsn claoisioii malcingo, Pliilosop]jtQrs ,hevo f ouncl no lojiloal basis, for oliolcQ betvvaan ths fceloologioal and mociiariistio raodes of In the explanaticiiQ Unit botii modes load to v;hat appear to bo unsnsx'jara'blQ questions « If a bakor bakes bread for ths pxxx-pose of SQlling it^ if he sells it for tli© pur- ano. p039 of raaking Bonej, ana ha of buying fooiS., v;antjs money for the p-UTposs clee.rly thia strin?^ of c^b joctives Qan bo OAtendGd o-aaily to a quastioa of ultiuats purpose e,t least cvirrently, impcasible to answer. forO;. It seems the3?e-> purpobao On %he other hand, is a balder 5 if © baker bakas bread bsceusa and hs is a baker because he choso to ba^ CODS one J ar^ be chose to beccsis on© beeauBS his father £1 1/ *" is, that teieologieal ssiplamatlons are bounded by our kno^iiledgo of ttitiiiiate iiQ vihlGJx bakor, '.vc find this string of sEplpnetions 3.eeA'XnQ i!?as ln(^T.oy:&hlj For a mora CG-mplste dlfjcussion of these terms se©: Bortrand Russell, A Kjstory c^ Ijestern PMiPS.^?]},^* Siraon and Schuster ^ 194dj J) o" HB"; i:nd Ec'^lTl BraifirA'aite So^lQ?:..tifJ.^R iiE^LS:^Jil3LS Gaabridge lTni^"ex^s5.ty Press, 19ijS7 ClTaptarTOo :.4. to a quGstion ot first cause v;hich ia also iniposaible to an.T.var* Thus meoha.'aistic eisplcnations are bounded bj our Kno-.yledge, of ult,.lmat8 parposSj not. bu-i; of original causOo Daspita tbo logical d5.saclvant-as33 of of eriplsnstion, find thao v;o nicst "ootji theory can be roughly catsgorlaed as eibbor teleological or nschanistic. ien mecbQsiics Ik. '^vhora nodes 1^0?:?t.on- tho aolJions of bcdios are oJtplainsd terms of prior properties like mass^ valocity, and tlio foi'oes acting on tliea is clearly meohaaistic. \^Ml9 eaono-^ mic theory tvliioli assurass baUavlor will accoiiiplish the pur- poao of utility aaxlaisation is obviously tsleologicalo lu Bcmo QaseB, tlieorisa in ©ech moda devisQd to ozpl&in.the samo phsnoaeriOBp light rsy through a leiL-B: systQEij for oxsaple, pleinsd both machaaistically in properties of tiis time required terrfis boon path of a can bs es:- of tho refractive systea and the cbaracteriatiGa of light, end toleologieally in terms of tlio Tiia l!.ave for- a Qlven dostisaationo tiio path vvliich v/ill miniiaisG light to get from q gi^en source to Kepler de-«'ised a teleological thooi'y of planetary motion 'which precaded Newton's msclianiGtic theory e' -Thus it appears possible, in principlSj at least, to diocovor dual theories; on-a raacliariistiG cal, by whicn sveiits can ba axplPiinod. and ono tblsologi- If this analysis 5.s correct, en attompfe to dooido in general which mods of theory is sppropriatQ to a givsu question is a fruitless oj:ie f?incQ ,Q. givon siiTt'iciono effort, thoorios in eitlier or A ooulcl presi?,mably be dovisado jsuch Isotli moclos cere appropriato qusoGilon porbaps is, givoa that no satisfactory theory exists, Vjliai; kind of uheory seecis nost appropriate to "bulXOl? to attfsaipt This question grents the possibility of lisoful theories in either- aoijropriat© critar"- eacl s^Agr^ests tiiC rrioda ion should be ona of efficiGncvo In tho malcirig, att6iiipt to da-'7isQ ttioories h-jsian decision taleologicel t'nsories are particularly appealingo By intrQspsotica most of U3 are are strongly affaoted D-iako of bj' ev7aro that tba goals viS tlie deciaions It sqshjs seeko appropriate therefor© to attsmpt to imaorstand deoiaion; in.g in torms of such goals o goals vihicli vvill b8- tiaaful TI15 v,'a snaJc- search for a g^noieal set of iu unclerstQiiaing tiio decision pi'CGoaa Las ^xad lisiited success, l20TfeYer<. To illustratQ tas process of a search for predictive f;;oals, coiisiaox- tlia prclilejn of d3¥is5-ng a ccd thGory which v.lll predict tiis toleologi- form of a body of watorv Moat observations indicate, that a body of i^star seeks to iuinialsQ the distanotj frora its esntsr of gravity to ths center of tfcs oljjecti^e irvill a theory bssad on iveter hai'ing this frequsntly aaka corifoct predictions „ Suppose on© r:sre to jislf fill a bucket i^ith v/ater and v/hiri hovi'ovar, Sj Qarth and — For nore ccinplote discussion of sosis of the issuoa of these tv^o raodss of theory ssa: C„ Go Hample and Po Or/osnhsim, "Tho Lor^ic of i5r.plenatioi:i". Philosoohy of Scieiica. ^— ^ «=-—-—»•' VoXo 15, 1946„ e, ' — Es would observe it suitaoly arovrnd his head„ Ijliat watoi' woiad stay in tha buckat at the top of tho arc sven though tlie diatanoQ fro:a its center of /gravity to the center of In this the earth is not rainirrdzed. casSj, tha tiioory fails To explain this e?ent one raight invent an ad for weter in i?;hi'i'iins buckets or, if he Iioc c, objective tiiougitt of it, SGHQFalise the purpose attributed to v/ater as one of r-iini- mising its potential energy subjact to constraints o This latter theory v;ouid explain both piiesiomeaa in terms of tha sQCje goal ami '.Toxild. as a reoiij.t. he a aors pov/afful on© than the original, qs long as the conoapts of potential snorgy and tho constraints v/oro operationally raQaaiirablQo Taleoiogloiil theories of deoision raaking hav© ariGou/itared r>na3.ogov.s probleraso >milQ In situations of car-tainty people decide to chooss tho hishor of ti/o alterna- tive amour.ta of raonsy, in uncertain ones they fyaqriently don't chooce the maximusi esipocted val-aOo .imi?.ins Tco o:qjlciln thicij goal attributed to psopla in this aituatioxi haa bean modified in two ivayso in ohQj another ad hoe goal of uncertaiaty -aToidsnce is offered Y»hich is in the sscis spirit Rs an ad hoc theory of v.'ater in v.'hirling buekstSo In the other, the iadividual's goals ere ganoralised to a aeaexire called titility ivhich is similar in kind to tho geiiei-alisation to potential onergy in the water cesoo chief disadvantaga of utility as a pradictive deviCQ is The the yias- •!«,-, -n nr oporaoionailg' cMi-^-iic- it.."' a.s>.o jiiSiiSUXiAg, -!-^. li'-n -s.^-' ptt^Gm•ot. >-v.w„-_,. to Burins tho psst so^aral ^©oxs tharo to^3 arioan o-?i c^.o 80 con- IiaB ©aiargod of a faaling of alSBatlsfaotion T'dtii prosroos tlieoriQS of doaieioa and premise in tho classioei toleolosicai firsli.tiiae of a isr&aod» makins and ttio e:w'xl.lahllit7 fox tha Tajs v#ork deRcribQa aM propog?©d la tliis paper Is intcnasd lasoliaaistio tHe^^JTo? to oont^ibutQ to ^he g^^oi^las bo^y of si.i>ii.ji.ii«i-jii ^ aM^^Si^iSal S3i.noea% I' 'ii»" i^-TPii&'g m^m^ lESSSd^ S2li^^ Tolo XLIJ^, Jun© 1959 J ppo 2e&-abbo He Ho Cyart and Jo G. mvGh, Ml§SiS^^ Si£C£Z Pyentica Hail, fortiioomj.xig i9&o. i~i:>?:a. ^ B® ^•e^ it Hechanist.ic ."The orv; of Goal F^H^^^^ the concept of Most mechanistic t.aoorles find amount of evi.v substaniiial u.Bnec9ssaryo purpose to be and thsr^afora human dence siisgosi;3 tbBt human bshavior, =^hen ono attempts a decision making, is goal oriented. bo mast therefora mecbanistic theory of decision making, fashion esplain tbe eitbor deny tbis avifioiice ox in some in terms of concept of a goel aecbanisticaiiy, ioQo, chosen herso prior events. The latter course is organisra is capable If ws aaaviac that tbo himan of the total information of receiving only a small sample further that it is SYS liable from the ea^ironmsnt , and only a sa^all fraction capable of prouessing (ettandlng to) receive, and that this of the inforsiation %7hich it can deciaion maxing, procossod inforn^ation is the basis for we nesd a theory of then to understand decision making capacity is aliohow the limited inforiaaticn processing how the. limited receivcated to received information and tha e'^/ailable information. ing capacity ia aliocatsd to two coding processes The hypothoaia advanced h'sra is that fratasv/ork for aneh a theory. ca;i provide at least tho of eategorisa= Tho first coding procoss i» that the decision maker codes tion. This is a process ^.hereby be called variable the raw data h8 recei^so into what might -9- classes. For e:Earaple, an oxacutlve in a firn attends to a sat of reports on such classes of vareiablos as share labor cost per unit, of raarlcot, force siae, forectist These vai-iabls classes contein far loss dainand, etCo= injformation vjorl: tlian. is availsble in his anvircnaent end as such his system of categorias conctitutos v.'hich a filtGi' he coritrois the iiiforaiatioii ha recaiveso aiaplss can be by Other ex-" constructed vihloh suggest that the categori- gation process is a canerel filtoring; raothod not confined to oxeoutix'os or sianegsrial behaYioro The process by v;hich eategorios ere constructed is o.escribed in ths psychologicel literature under the gsnevral title of concept forxiiation 2nd baa for the most part bsen studied in the context of a labox-atory situation vvhsre other v,aTts of ths decision ';rocftfis v,-are largely It 13 suggsstsd, however, that the theory supprcissodc beginning to smsrgo from thet vroric will fit into a general theory of decision making aa the fir&t of tho tv;o coding processes bsing dascribsd ha.re„ ^i-ssiming that is codod into v^ information from the environiaent set of meaauras on a woll defined (but not necessarily constant) sst of variable classes, tha next problem facing ths decicioti aiaker is to allocctQ his limited 5/ For a description of this woi-k see: Erimer, Gcodnov/, and Austin, ^ Stud^ oi; Yhi'?king > V/iley, 1956^ -10- prccessins abilities over this set of variables o It is by tho suggested that this allocoticn is accoaiplisliad ona by v.-hich YGri== second coding process. This process is by moans of ables classes ars coded into t^^o catogorios values called goals iiaasuras 'f » en' variablo fflessas ers constontly ro- workfo:fC9 ceived froa the em^ironeQnt, e.go, nien, estimated shera of market = 62;^^, oto, si?.o - 100 Theae measures eppropriete variable aro GOEipared v.dth goal values on the a aimplo class and, at least as a first approximation, results. If a meaisuro excoe-ds its goal categorization dofinition of tho valuo (where or.oeod is dof inad in the member of tho oot Yariablo olaaa) this variable class is a no further of variable cl^^ssas which requires processinso If > lasaaura received, on x^ha infomation other hand, a goal exceeds th8 the this vrariabie class is a vi.ember of oonstitiites problj^, and set of variable classso v;hic.a roquiras information processing of the probleio. solving reduce th3 diffor^ioao. find q bshaYicr routino to 6/ The execution of goal. tho aiid msasura tha boUi:'een once type constitutes the rcutino discoYored by this processing observsble behavior 6/ Solving GaneralProblera A. Nev;all and H. A. Simon, "A arid i^iiiOiwti^n, Coiapu-oers Program -^or a Coiiputor", Svly li'60, poo lO-l^o .'o... w, .==11 variables Tho existence of goals on each class of processing called d09s not guaranteo that the information to the information for v.'ill bGir any particular resemblance particular processing capacity of t.hs decision makoro A idla tha satting of goal values could overloacl or leave will problem solving machanism, Thssa tv/o possibilities be considered separately^ such Assume for the moia«nt that goal settings are that desiaiida for iD^orKation processing exceed the decision maker's capacity to do tais v^orko viag thiis Tvjo raaohanisraa typo of allocation probiea exist „ for sol^ Ths first is thereby remain to ignoi'Q one or more variable classes and ignorant of tha fact that capacity is e:^ce©dad. second is to modify goals do'^nv.-ard The in such a way as to of ths bring the numbor cf probloms within the capacity procossins syatem„ It is fmggestsd that both tiethoda of reaolving this problsm may b^ usad. processAssuming convorsaly that tho information at a given goal ins capacity is not being fully utili'^sd to the decision asttins, two alternatives arc also opan variables to maker, lie can either enlarge tho numbsr of goals on the existing \7hi^h ho will attend or he can raise classes. may be Heive, usedc- too, it is suggested that both mechanisms ' Willie the J. w four mechanismf. Just dQScrlbed v/ould accomplish an alloGatlon of processing cepability, these clQsoriptions offer no ;lnsii;ht into the process by v;hlch these accomplish this allocationo la'eGhanlsris It iSj therefore, to tho problem of goal modification and th6 probloiJi of controllirig. tho set of variables classoa to vvhioh WQ will now turno Asstiaing as a first approximation that thQ in- formation recaivins avA procasaing system ia aeaiisntial, a single sequential nrocess must bo cspablo of performing the work of the four tnechanisas Just describado ViTailo z/ it is uncniostionabiy trua that the cat- Qgcrisaticn process d'oscribed earlier is carried on in con.juncticn M'ith tag modiricaticn of goals and '.'ariebls sets, it will be assumed for aimplicity that the total set of variables olassss i3 gi-zen and will remain fixed, Tho follcv;ing theory is sviggested for the wQchaniEius dascrib^sd above „ Soe Fi,gure 1, The of all variable classes and Qssocisted Slit goal values ia stored in a raeinory dsv'ioe. The decision maker has a priority scheae v/hich ordora tho aoquanco in 2/ a ganeral dieicuasion of thsse E'ochaniarus and their IJarch and Siiiiionj ituplioations for bohavior aog; ?o:c Orfjanizfitions . V/ileyj 1559 -13' SriT 1 r oniae n1; '^ 7K 3: Coiicopt StructiAre HeaeurQS on Variable C1SS3C3 ?r5.orlt7 Hiilci y/- Head Meaaure on lMe:st Variable (as defined Dj priority -iaile). Store — Goaparo Loaaur© >? Obsorvable Bshavior T,'ith ,Goal for this Variable Class -^ / Satisractory Unsatiafactory ^.\ Goal cation MocUri T ! -14 which hQ vrlll p/ctond to -the varioiis variable e.lossgs„ Schemes such as a series of reports at different frequQn:ies like weekly labor reports, daily accident reports, and annual subaidiary are suggestive of formal procedures reportjSol' a used to ecccmpllfsh this soqv:6ucing.-, Cnco a measure on a variable class is received, it is compared with the stored goal valusj Tor that class o If the porforraance is unsetig--- factorvj problem solving ecstivity is baguno nhen this processing is complete, an action prcgrac may have result^ ad and exocutlon of this prcgraia is undertEken, may require scmg time to discover tha effect 03T Since it this action prograia, the goal for this x^ariable class is modified in accordancs with the outoorAQ predicted hv the prob3.em solving routine and is stored Ono can raaho foi' sorfte tho nest reading on perfcrsance, general propositions about the parformanco of such a routine^. For Qicamplej one xvciHd expQCt the the loop invo3„ving tini^ rsqi>i.r-3d to' Q3:^^cute. satisfactory pox'f orraanee to roquiro loss tiiaa than the loop involvins unsatisfactory pex'formaneo and the concomsiitarit iaent problem solving ^'outinoSo yislda maafmrsa values, on^d virhich would expect it Thus ^vhsn the environ^ excesd thair associated goal v^'ould he pOKsible to get farther down tha priority list of var labia clasaes than \Ajculd bo poasiblo ir the enviroiument ..oro Tha priority belOY. goal values. providing raoaaures nao for controlling tiie would, depending upon the next variable to oe consiaered classes which environment, detern^ine the sot of variable will 'ue evaluated the decision It is interesting to note that environaiantal himself detertriines v;hethar the associated goal valuos neasures er© above or 'Delow the rulOo If, ^"or through his choica of a goa}. modification arast-^ v^hich is satisfactory ^mder the Eialcar 3:i:a:nple, a moa3Ui:e aharply raised hof ore ing soal reoulte in the goal being that ovor tima this it is stored, it is quite iiksly thus constitute a Yarlahlo will heco^e unsatisfectory and cT-iniilarly, if the goal associated problera in the fiituro. currsntJ.y a prohle-ai is •with a '/ariahle clsas vjhich is action is sharply lowered vihen oorrecti-fa this action ^/vill ii.nderte.ken, alaoat surely ha successf-a.1. Thus the ..ith the priority pxo» goal modincatio2i acheme interacto the set of variables cess and the enviroruaant in controlling attend and the amount to t^hioh the decision malcor will will undertake. of problesi solving activity he have heen dGscribed To Bui^mar-iso, threa procsssas the infomation the v^hich, acting together w3Jj. control his limited problem decision n^aker tscoItog and allocato -15= solving ability to a limited n\iK.bor of problem verlableso The first of tlioae procassoa ia that of categorisiationo The second I3 a rjiiority assigiiiaent process, end the third is a goal nodificatiou process If thesa three procesaas can ba supported by onpirical ovidenes a long step %7ill have been taken malcinsc goals are to-wa:;:'cl a mechanistic theory of human dooision Indeed, the thoory dsscr-lbsfi above sxis^^esta that detei'SiiinQd bjr properties of th© decision maker and his en-^viromr^ent and that teleological ass umptions ara not nocossary for an underatandirg of goal oriented human bahsTiorc '17- mechanisms In carrying out research into the first cleGcribed abova, the resoarchsr must sat of variable clase.ea to ^vhlch ha v='ill datermna a attend and then oonstruct some goals Tor his invest igatl on o It Is oon« into these processes coivablo, therefore, that the research resoerch itself would could provide the data on ^vhich the bs hasedo wiiil© The properties of such an undertakiGg, presant at least to fascinating to contemplate, appear at a rather of success o be too complax to yield much hope A set of \^Gri = mora prosaic approach is plannsd insteado but the processes abloR Y;ill ha chosen and goals defined, v/ill bs left undefined. of thsse choicoB and dafinitions Three basic processes have bean dQscribad: modification. categorisation, priority assigrH^ont, and goal last of theso thrao T'hlB study mil be directed at the This decision is made larga^ processea, goal modification. is tho most appropriate of ly because it appears that this araa a ainglo variable tha three to study because it requires two processes require claa« to operate, ^vhereas tha other appears thst aa two or moro variable classes, /dso it will be a understanding of goal modification processes t.70 prccesBes. to an underetanding of the other proi^oquieito .18-' study has of efficiency, tMs Thus largely for reasons modification aroa^ been limited to the goal v;ili ^e ooncarnQcl Within tbis area the study performance on goals, not of the only v^ith the effect of variables on goals o WMle not effect of othor Kinda of it is suggested desoribod in the section on theory, independent . On the conthat vax-isble classes 'are not to ass-oine the general process trary, it aeema reasonable hierarchical ao that variable of categorization ia lovela of abstraction from clasaea are defined at varioua goal information. One would e.^poct the rav; am^iromasntal classes at all levels values to be assisnea to variable on one .^ a raault, goal modification of this hierarchyo either from information class of variables might result information about another about that variable class or -.^^^;!.^^ are related hierj^Ti'-i "o'-33 ^oj.j. t^na variable ciase vvhoai'' aeiino-ciou and 6oal in question. arohicellj id.th the variable class feedback of iBXori.ation on the Goal modification' due to defined v.ill do defined variable cla.s for v.hich it is Goal ruodifica^ on the gcale ae the effect of performance other variable classas ^ill ticE due to information about en the h^fivprchical V -''•^•<-r.^ aiorari..uj.v.'a„ strvicture bo defined as -Ghe ei.j,oct o.•;--, , r,'-. r,-^- goal .: ^ ^,« <->>', c^ '-ifuA^-' the effect of perforn-- •u ^ ^* ^v> ^T^f-T-ast o of int^^xas^o sub.iect anc8 on goals will be tne '^hia choice =19' problem purpose of simplifying the is nade largely for the studied in the context of a single into one ..^nich can be variable classo Y.'hile this study is the domain of interest of theory to which it propose, considerably narro^ver than the this simplification is a to contribute, it is felt that oontr-ibution of thia study to realistic one and that the be a valuablQ one. tho mora gansral theory vdil ^£ lii^teratu;^ or theoretical Thore has been littlo eapirical modification sub.jact of goals or their v^^ork done on the branches of acionca noat con^ through experience. In those economics and psychology, the corned ^vith decision mekins, for tho most part assumad goals of tha decision nakar are classical oconomic theory this to ba fixed and known. _^In theory explicit r-" In much of psychological a33VUP.ption is explicitly but laado none the leas. thi. aasi^ption ia mado loss classical (tsloologicel) The behavioral basis of from ^hich the reoxainder economics is a aet of asau^ption3 Difforencos of opinion continue » of the theory is deduced by ^hich thi. sot of exio-ois to bo heard as to tho criteria empirical tests to vhich should ba selected and aa to tha 8/ ^ J. n Ho Henderson ana R- -. iSo — -—— Tboory,, Avtoflconomic •.,r^-r,r-!+ ~^v-uandt, &iiS.iidi!_—_^ kcGravriiill, Kev; Torlc 1958. ?, .5 '20= concern of classical economiats they s)ma.d bo subjoctr"^ The relevant to ttris study. for goals tliougli real, is not of econoi:aie beha^'ior R^DCsntly a nochanistic theory has been sugse3te.r"^n v;ln.ich goals of the typo dascribod important part of the theoretl^^ in an earlier section forsi an work on goal formation pro^ cal structure , Little empirical and the authors of that ceas has been carried cut, i^cy^mver, the need for a study of theory hava o.^:plicitly recosni^ed described in this paper has largethis phouoracnon. The study theory f^nd takes as ono of ly accepted the structure of this that its objectives a ooatribution to vvorlCc psycfeolosists ha^^e not bs-sn directly inter- ested in the processes of goal inodif ioetiorio for the most part utilised tivo They have valued re^vard systearH punishment. vrlth vast differencoE betir/een re-ii'ard and electric ahoclLi5, huagar, etc. as punishiaant; and food, lOTe, lack of shock, etCo ea rewards. 9/ Oogc-s With such eztrems effects Hl L^'lJiJ^ilt F^r one Dosition see: Uo Frisdrsan, Sssa^_ x. J9a3,^Caapuer Chicago^Press, EconcgicG, University of For enoth ln£ of __ FcVthcoSing, ^ _„ 196'o, Chapter do H. evert ana J. G, March, Behavi£ISi I^M^SZ 2l prentice-Hall, forthcoming, 195i5, )a/ Ro ^^ 2HS* 21^ the assumption of the goal state vjhich the decision ma-cer seek<? to accomplish is reasoriab-ly cloar^at least to »jade in psychologists end no explicit iittcmpt has beer their work allow fox errors in this re?>sct.. ;;'^ goals TJiiis have not beon explicitly deelt with \^ psvcliology any more than in economics.-, Thore is one . artia or tho psyoholosicci litera- tura which is relevant to a study of goal modification even though that vms not its intent In 1950, Hoppe sufigeated a nev; Drooedure for tfiining a behavioral measure of parsonalityo ure v;as His proced- carried out by Dembo in 1931 end his uioosure givsn the naraa in, a laboratory situation v:a.s This nama was levsl of Bspix-aticg.„ strongly suggested by the iV^xparimontol procedure: subject ob-- a asked to perforn a is sequenca of similar tssks w/isre his psrfornance can he moesured along a simple 10/' F- H'JDpa /'iCrf olg und par:?ors:isnce scale i'.iiaiisrf oI^k, " „ Q = G- , timSp pg yp,-^ol.pg,i g cho £2i^.2ilH£ia> 1S20, Vol .44, pp 1-5 a. 3J.y T. Dorabo, ISSi/vDar £?/,2Skol-Osi-.siIil§. Jirge.r als Dynaniisohes ProelaaiSj" For.3oh ung 15, 1-144., 'ZZ^ distance, nuabar or errors, etCo task he is eivsn a raaasuro i^toT coHiplGting eacH of his pen'ormance (score) performancQ he proposes and aslcQd to state tho levol of The score to which he bo accomplish on tho next tasK„ his level of aspira^-^ aoplres on the next tasv is dof ined as va-lters siuce have tion for that task. Decibo and all level of psr-formance defined loTal of aspiration as "the vnieTfamiliar tssk ^A^hich an individual explicitly in a takes to reach'' is Despite tha fact that this definition the word ally identical to that implied for virtu^-^ £^ in despite tho fact that earlier sections of this paper, and in ressarch into the phonoraancn is reported e ;terisivQ directly useful the litoratura, this literature is not modif icatioa."-^ to ono interested in the i^rocess of seal aspiration hava boon Since those interested in level of dofining a measure primarily concerned vdth the problem of the rosults of thoir of personality, they hava tranaformad experimants oa each suhjact into 1 V "' a single statistic \^hich of niay ha uaefva in tasting a theory '^uSea p. hov;ovar. s'oal moaification, It the procass of '23" 13/ agsregatod tho task-to^task beliayior„ Tliis statistic its conaistanoy studied across tasks and subjocts and under various inwith other personality raaesures and the prasent study fluences- WS.S noted, Unf ortunatoly for v;a5 and aspiration level th« task^-to-task data on perforaanca is enti3.'aly missing in tlie literature shows Tlieoreticaily as wall, the literature Kodificationo 'S^o little concern for tho £roc6ss of goal But the authors address themselves to tho problemo neithor an operationtheories proposed ere toleological end suggostod to ba rele-^ant, al means of maasurins the Varieblea IS/ Tho particular statistic usod n v?he.r6 v;a3 the follovdngo '-'^ D - average dirfereaco score - total numbsr of trxexs n - trial numbs n H ^"1 = Esoiratioa level for cnal n - pert onsance level for trial ^ . Is' ,„,„.,x -- Pj^ ^A/ "Level of Asoiration and Dsciaion Making", ilk^i^ksi Hevie^, 64, 1957, pp 253=^252. <-- siQ^el nor a Kiigbt theory cf the process MhQTGh'j ths subject th& prasoriiDaa e^'aiuations is GuggsstQcl. c^rry ou-ii tliose theories of process v/bich are clesGriDed in Tiius tiie to litsratura on laTel of aspiration are not relevant this inquiry Several authors obser"?"Qd almost in passing gucgsss and that aspiration le-vols tended to rise on fall on failur©o 'X'hese observations appear -co consul- goal modification. tuts the state of our smpirical kno-.vl6dg9 of 15/ ^' Jo Do _SranK:,^ of this "work ses: ^ Aspiration", P£S2£Saoei' ''HocGjit Stucli©3 of LQ-?el of cal Bxaieti-aj 1941, pp.. SlB-225o ?or a s^3JB2•a^y 85^ «»£i»i«*' In vie?; of the lack of appropriate dalia on 'Gli© proccidure has goal inofiification procose., en exosr-issntal boen dsaigviea to flolA of this proesdura, aaaig^i -;>jill t^he auov- data. Follc^lng a daacriptioa consitler-ations vmioh led to this ba aiscuBsedo AftQV roadlng a detailed set of instruc- the perforciance he tiona'^a sulDJeat is asked to indicate hopos to GOhie^Q on iadioated in ]iis tlia first cf a se-iea of trials. i.s inotructioas tha porforsianca laaasuros, Ha indicates, are in ana th^5i*Qfcr6 the goal I'alua which ais firat goal units o^ dollars and corrts, Ea writes trial Yaltie in a blanlc opposite Yided for tuo purposeo xnmbor ona on a foria pro-. Tho sspariffisnter then compares nunbs? ens vdth a ta© subject's goal value for trial ooasure£.„ If thQ preYicualy prepared list of performance tho aubjoot exceeds perfcrnianco Ysiue ha tells If the perfonsance that h6 -lost^' o^ the first trials tho Buhjoct that ha value exceeds the goal value he tolls If the subject ^von he ciroles '^woa- on the first trial. ^ritea a number ona beside the goal TQluo on tho form and casa to Ha thsa procseaa is oithor the circled valuo. P-oal ?al-ao two, and the px-ocessa inclicato his goal for trial ntimber continue So ** Apponais Ao trial and Uaknoi^ to the suoject the cost psr the parar-atei's of BBleotad t.o porforaance data diatrubutlon tiia i.are game quite benign, ioSo, a wi^a range laako tlie net oarninss ovar the of goal values viouid yield poaitiva course of tne QJsperinioato Witl^in this constraint a variety used in the er.pa?imsnt of trial costs and parameter's vmre if any, of aucT. changes on the goal to reveal tlaa effect/ mcaiflcaticn process <, believed to FiSUX-6 So ind.iGates the r-slationship and tho theory esist batv^Qsn this ©^iperimantal procsdurs and rsprossnted by Flgur© lo i^ig» disoussea on pages 8^a6 ura 1. for convenience, tli3 oTor is reproauoecL in Figurs So ^lie tfao title "Thoovr otHer cliaerara in Figure a„ represents e^^parimQiit o aams process as it ^ould apply to ths instructions Concept structure is provided Ir^ tha only one -oriablo and tae priority rule is trivial aincQ is relevant to psrfomance. Problem solving Glae.s ia kBo^a to be predster^trivial sinoa tha perf orsianoc data is tlis only part TbuB tfo^ ecal aodification routine is luin^^a. of tha ps=aoe0s i?vhie^ tlio subject must consider. ^^© espori- p-ocoas and oaly thia pro^ K9nt thoref ac8 -iolds data on thie oe.'is situatioiio in aa otiiex-wise vex'y structureu several clioicos mada in tho design of lao^al i.ituation deacrva ^irx-IofsQ inforiaation coi=n:eiito Tho feedbaclc of siKiplo on psrfcrmanc6 altoxnativ33 availablOo It ttcs tlie 0>'p0ri" x^as oiLl.y one of sovoral aalectod prinoipelly because to of tlio difficxilty in so simpls a situation simuate ' -^ 'ZiU-= o ^^ rS •r' Q o o »=•<;;*'= ©ffect o:C vt=!luQ lu io expected that en performancoo The choice of th© variablaa ors not incleponcient » tlisGQ tv;o vrln-3.03e goal msasurs asaunies they are ]iig}iiy depQiidento It assuKos in fact that parfox-mance is Ximited to the goel veiue on auccoas and lii^dtod q fise-i increaiont helavi the goal value on failure o The size of this fixed increment ia set by the cost per trial o sul"ijsct '-.rlTkB SincQ t]i«* er-cpsrinsnt ccntlniiso lontil tha a fixed nuabei' of tiaea, there are no "opportuni- costs asaociated vvith a losso ty"" Only the x'ixsd cost psr trial is rQlaTer.t to this casQo Tho fixed cost par triTil sign of the In 0. G?^.pcriiffiant ^tis included in the dQ- to sim-uiato opportujiity costs, howavoPo niore co.ijplos sifci-.ation the sub^loct v-'O-ald he forced to accept opportunity cost a with respect to. other- vcris^ble oIqssos each tims he tJ-loeoted his attention to this variahl© class ,. One 7;ould oxpcct for exfiffipls, thst with graat dsiaands for prohleiji solving capacity, gosis on individual \'erlabla The classes aight tend to ho lowered to redues tho dsmando cost psr trial pcrraits explicit variation of this effect c. The cyltorion chcGen for the length of tho e^sparilaent was related to tho cost par trial considerations o If for OAauiplc, the expsrimant vjere defined to last a fixed munher of trials, tho total trial cost would be fisad end tho offset of this parameter of the experiment would ello-iving tha ha'FtJ been lost,, Qsperiiasnt to isst for a flscod number of Ey "v-.'ins"' total trials bacome a variable and the "pressure of other ¥ariabls classes on gcale boecmss rsalo This choice 6\leo olim- inated ths .opportunity coat v/hich othcrsvise x'vculd have been .30» perfon^^^nee data InfoTolns tns 3Ul^Jaot that his saquon.o v.-as based on a few trial .vere arranged in a random these trial e:.pariraants tlae rune '.-here t-Ms was not donso In seelcing baUavior in subjects initially engaged in pattorn tUat noted in a study of the tHa win^lcse sequanca not unlike after e mx^ov of trials did hlmxv^r cboioa situation;"-^' only hypotbesis the-t tbese patterns ex« tiio subjocta reject their conceptB in their goal iBted and begin to use more General. ....^u the osquenco •r~,-«o->-^vn- the 3uujt>t.«o "^ub-iQCts that t modification proeasso ImOi'mins ^xie elis^inatad tbis early of perforsiance data was random largely to tbe experimental behe^iox wbicb seemed to be more relatea processeso situation tban to geiieral goal modification benign situations The deoision to study essentially nost of v.o require tbese for was based on tlio ass^^ption that subjects' pro^ Xt vje.s felt therefore, that tbe oui' survival o more at bosie in tbese cesaes for sosl nodifioation might be Xt v^as also iii^iaineat. situations tbaa in ones vibere ruin is ruin barrier on bebased on tbe belief tbat tbe effect of £ dspsrture from bo^ havior 1^7 OS most usefiaiy studied e« e, an indepoMent process „ ba-^ior in e, benisri situation tban as a useful oneo The e2iparin-isntal situation seems to be nmabor of dlKansions. It S9n» It permits easy variation along e easily. This study has urates trial by trisl data quickly and considered only a smeai sample of tiio potential situations vvhich can bo created -sitbin tha gsneral frarsswork. "^ },»• / Jo ?©vixvaTio 'Veil Aaa?Lysis of iTedictiva Behavior ^ot.ChSic6 Situation^- uapublisbed Institute of Tcobnolcgy, 19b<Jo ifh.T). ma tboais, Carnegie "Si- To dat.o tliirt.y tureo Bubjocto i).a"ro participated sxperiiuant. T..o o? the., .ubj.ot. in variations of the basic a total of tUlrt^y fifo Have payticipatad ti7iCQ siaidng Since eaoi. azperinis.^ genar^ of tba experiment. roT^etltion. ate. al.o..t the goal modification. ona >mBd.ed observations of yielded ootween three and proooB., those e.:peTiment;s ha^a The auh;1ect^s ha'?6 heen ro^^ thousnnd such oDserratior... diversity of hacvgrounda. Thir=^ ara^n from an Intora.tins Sloan Feno..s in the tean of the subjects are currently They are, thorefcre, ex^ school of latotrial Managemant^ in their ^r.---oet?anJY viewed r.s sticceogful -i«o r>-"^-ir-Tvr-P caiears at aga 55 to 40. Thirteen of tha subjects ar^ curr^ t^o are KioDo candidates; ontly ma.tar'3 degree canaidateB; -^^-.^ r^^ t^'^ fHcr.'>t^ at the School cf Indust" rial Manage^eBte repr6seutati-.o While this sample is hardly of the popiaation at large, businsas a^GOutive it Eiay ao^v ^Biorgins represent the class of and malcing on impact m the profession of •^aaiiagemontc "h«t tha purpo^^ of this It 8hmilfi he emphaBi^^ft by .ioh s.t a^ <^- (tri.l .^.ariiaeBt is to £.norata a mi^t he po..^= protoa.lB) frc^ ..hich it .lus hoha.ior t.i.3 moaifi^^ . theory ot goal oonetruatio. the start to ihle oh.o..ation of ho ee.^ea by the can objective Thio cationo for o'.on such a Bv3.^ect in a uU^&Le experiment, ^ a .In^O-o small set of data is riah ^ o:aoiign to prO.iao .a. xr.„ :Ccr a thooi'y of tho A ceooud sv.bjocb will modify behavioro Eopefuily, that theory scniaviiet and so fortho fron bafca-g-ior v;ill be JLndieatiiro of Cn3 ssoka e sot of gener-al structurQ which t!i-3 <3.et.a, thex'afors, sup^rAiciall^r aiffor-ent Ijohavior v;hica can a saall set of undsriyin'g processou. "bs -Kjhich exhibits ozplainsd by In glancing at the data a oompietQly satioAactory sEicimt of div^ fie2ieratsd to date, Qrsitv' is ©Tidorit = an nnovmt which thaory q rat-lior kdisII sample ot snoh. cosples: K?liioh esj-ergas ia sought o tlie -^vill e2s:plaia 7:?aioii suggssta that a thaory a nignifioant samplo ot this datia must contain oonsidsrablQ gonQralitj-q Essic'.se which hibits ths usoAiunass of the ts?iQi by triol data hg dirjGussed in taa nost sectioaj the data ex- T.'ill sQTTxo intsr'astiiig aggregate propsrtisac 0220 of thase hQD to do vjith vjhat might ha Gell<5d oasio strategies in the gool raodif ieatioa prosesso A ssiail fraction of tha suh.jsots undortook aooncmic analysis of tho exporimoatal sitiiationo tioal criteriori h^i- l^ioy oomputed p. statis' which thsy oould econoraioally ©Taluats thsii' 0j:po5?i9ncQ at difrorsnt goal lo'yolSo 'rheir hoha^ior than con- siatod of gathariiis the data indicated hy this caloulation^ Ovor ths coursQ of a hundred trials ox so thay sTJCcoeded in diaoovering a goal vaiuo vmich aet earnings vjciLl.d hcs close to thatAvhsro c:;:psstQd raaximi so d^ given full inforraation ahcut the parformarioe dictributioric thercjzisl^os i7as i^-oias that thsj? hed found the of these auhjscts satisfiod "optirr,^Jia" boforcs the ©nd of 'OC^ tliQ e:sparinent avA oftoxod to ocntinuo at that goal level for tho x-eaaindGr of thQ gam© Y;3.thout 'orial by triel 5.a?orsia~ tion on perfox-iaancSo A lorgar rrGCtlo2i of the sulsjeots used c sligholy clifr^ront stra'ijesyo Thay seemsci to apply tlis coneopt of ex- pQctQci veluQ uO tha v;in-loas record at veylous Iq'S'qIs but not apply aconoiiiiG analysis to this datQo triol reletive to tendocl tbe tli© ievQl of tlioir goal x^alue tiis totolerete mora losses than idns at co.'5t If cost v/ss pas? "Iovj" they a goal value o If par trial vma rQlatively "espQusj/va'^ hcive^er, they tondstl to choof^e cioal valiios ozcQQciQd lc5aQo„ in the "right QJhls vj5ie5?e the Q;:pactQd nusbar of wins slmplQ rula tonciod to bias their bohaYior fiiroctionp ioSoj tovjard the optiisuia, but they "* RjfstGpjaticslly under© stXEiatQd tlie oconcmic offGOts aiici rssult chose goal 'Values botwean ths fasdien and ths valiio ciifi as a optiLUiBi on the per-formaiic© distrilJutioEo Til© ^eaaining suojactr. uasd naitbor tha Qoricopt of eGoaoirdc analysis ffioaifieiQ 2205? t^io conGopt of sxpcjctsd a'-aluSo Ivacjy thsir goal ireluo on virtv-olly overy trial dopendiag upon thoir prior o.^perienog - ospeciaj.ly tlieir x^ocent prior ospayigncQo iBore of T}i3y than loosing tliQ uiiJ.i!cs tenasa to raise taoir goals vaon tvinnins enci Yica psrfosi-ueiico data, G 3.'and.cs3 v;alko -v^ereao thay followed « very erratic path not Thoir prooass led tliem to tend to choso goal \"araos In tha nQisli'Doysiood of districutiojio TliQ Bacausa of tha randomnoss ran<2o:-.a t].is rsiocliGn of tlie parf oraance appsarance ofthsir behavior , it -ma^'j^u shoiiia bG esphQsised, v;a3 <x\m not; process, to ' as vaXl ba discussea in t.he noxt sacticn, but tiiQ fact. elimiiic?.GQ a of ov'.ojoots significant 3.3 V;as quite "«atisf:lea" tliC5 tiio riLi.o setisfiod by dofinition for v;aa by his performanoe in ths is^itli -vjlrlch aatisfGCtioa was cisfiaedo eSiporiKiont each aubjoot was aska^i to cshoose the ainglQ goal value ha \7oiild uee if ho i^oro to reps at the exparisiQat without tho po-yer to ehangs his This is not surpri.'?lGg for \-?hich oash hafi used tlioir c;vn prooass of tho espsririonto goal modific«tioa tiio iov;c 'I'lioy bad, in off cot. built Qriterion on the porforaiancQ data d.ef inerl oncl choss definod Iiichsr levels aa too liafi high end iover lavels as too Ss?.Qh go(3ilo values olose to t'losa they had fourid neor tho prooasa not) intersa-alng to note tbet eaoli of tliosQ typos contrcl of ond of t'liQ c;^!© cf tho noiss in tUa porforaanc© amoun-'G In e sens© each QspQriE^sait 6 At noaificat-icn routins aid t,hat tlisir goaD, It ©acli kacl a ran^cm goal modificetion 'co ancl liad in the a ^Bl^-'ISSMSS.' optinnmo Ths fact that this optiEi'am in the vast ma^oritv of cssoa If vi^a not near ths uooncmio tlao G\.'t)jsct has no Ksans by bRGlc fi'o.m b^liaTicr at the "good'' lie position opti-Piui'i ivliich to racogniSG that tto feod- oconcmie optiemm is in sono sens© ob'^iotisly has no metina by v;hich ?7liiG3i is baside tlia point b_i£ Traluo systQia to profer it to a dofines as "good"o It \;'ould appear from these data that tha sconcmic optiHUin vdll look attrsctivo only to poople u'ith the concepts of aconoaic analysis in their goal raodification ijrocsGGo 'i*his result hcs acaia intoresting inplications vrluh rospaot. to attamptis to apply eeonoaic aaalysia to industrial operationa. Unless 'oiiQ cianager's goal modification pyocees inolucLoB tto appropriate conceptQ he migh-fc very \c0ll clofiiia soJ-utiojio •** sjorG the ocoaomic optiisua. to bo a poor carafui eseird nation of tho data should per- mlt empliflcation of tliia points V/ith respsct t-o the three basic strategies, it is interesting to speculate about generality <, Thoso subjacta appliad tha coaaspt of espactecl value clearly based most v;lio of thai? oalculaticns on the assunaption of a stable porfornjanco This was qulta reasonable in tha oxporisasntal distributiouo sitUQticxi since this fact -was given in the instructions, but if bliis fact assumption r;ex'o not given, it sooias doubtful that this ^voulcl lis°«© hoen inade so confidantljo It would seem tharofcrs, that as ths assurance of stability is v«aakanad, the procsssea of goal modification of thoso ivho used expected valuos might tend to becoina siailcr to the procesaos used by subjoeta who clearly did not benefit froiQ the infor?istion about atebility. Thus cno might hypothasisa that trisl by trial modification might bo tha most gonsral of the three processes describod above o This hypothssis can be tostod in variations of the Qxperimonto this result If it turns out to be supported by tha evidence •tvould make the point about tho possible conflict betv/sen subjectivo and economic optiiD.a ovon more interesting and iiaportanto A variety of other gsneraliKations on the data seera possible o ^'or Qxaxiiple , tha data and strategies suggest -36' v'^^'r^ ,^-cii v/ith uncertsiaty la their tea poxnt of Interest frcm og wi --^refore 1 . .>u3\fc.^.o.u be could ^/ 6nvisoa5iisnt aad tboory of a;icc.a^5.u-y of via., of tHo ganeral bonavior suggest that, xa^c car iLanDsiQTUxl 1 -<--.ri^ -a c€" sl'^rdies RoGQiit empirical „^ ,^. ecoacmic sisnificaace „preeter S^et^t^er ho-^ a. oe laa, VQriabilit-7 in bonaYior B^odification of goal i§.^ ,^ -tudy o of uhx. ..tuaj of thl. results ^'a^ Dias. than .nua a .or t^is variation, explanation an n.y sugseet that a to date suggosts satherod data the of l,riet reviev, thaor-y o. .ay .e de.ivea fxo. a results iate.estins or • , _ r^.. gool moalfieatiion proaesses^ '37' """' la orao. to of individual sot. ... Bubaects^ a* «« f^aa^eata. dl«oxonc.a based on bahavlox -heo. These t>ieoxio3 dif. or then be corapared Tor can protocols plus Btrucv.re t-be basic become , '.ill will n .1MQ. rao aim-llariaes axi-oies. TLe ai««. goal modif lc«t;xon. of <> theory of p.ocaoa and per=.3t-3 o. .he ooBS,.ao«d .e poir'c can u...>u..nea .e ccr.iae.aa ..yond tMt point o.n L'.o..no.3 behavior, rnero arc a nu^oer . = -^rf brood o-.time of «a .,0 a«t.e- o. ^e^...-o- this Di-ooacluie rspreaanus tlio a«e.cK propcaod. ^ protoool of ».« behevior. end or.a sun.oo. ..is -b,eo. .na...od oaxa.^Uy,, baen ... ,. oBe «po.i.e.. u.ad a t. al .e .aa o„o wno -oeoauca aa.X,.l« .,, 0.0SO. f o. «.s .e.a.io. 1. .a. faXt and ...a.asy b, „.,X .odincation c.o.ca ..a .i.do. o. V.i» I«.:betbaaaaie...ounda.3.and. is act inportantoc , ^..^^ «.i. 0.3, us. .,elaU« inc..a.ns ...Jeot., S,.o= .U p.ocoasea - .ar.3 l..a daoioio. ^ ..ei. dcaorlbing ,0... goaxs seems clear thav. 'it seem again, e.Co - x. bottor, lo^var, higlier, .a..er t.an o. prior .oals .odiricatioas o. .esul. a.a ... Civen inrc^.ion constructed .alues. ...s.l. o. a so^^ence previous fanure (lose) ojf iiis x>.a-uj. .^ (vnV; or lv;xn; sueoca.^. about, tbe aU..lon oonsidar. . BU...or o. goal. t*.a B.b.5aot, ^ of wna. be xl>.a., bc.ia tbe on aBd pxiox. eKparlanoe m -58« inci'eiTioiit by which he modifies his latest goal value to establioh a f^cal for tho naxt, triaio The theory {ahovai in Appandix B) attempts to ainiiilate this process o It consists of a series of questions enswQVS regarding pest experience and dopending on the an inoramorit is ohoson by v;hich the tjoal 13 to Yja Kodifiedn To disoorer tho theoretically pradieted goal on a giVGii trial, ono starts with the goal \'elue on the previous trial end knoy/ledge of the succoss and failure ospor-ience on ell prior Ono then enters the net in ii^ppendix Bo goal valuos» v</ith the result on tha latest trial end proceeds down the appropriate chain of quastions and ansv/srs mitil Application of this oradicted wodif loetioz?.o ividioat.oH t-b© iiiodifi cation declaratiVQ sentoncs a to tho latoat goal -t&Iv.q • yields the goal valuo prodictod for tha noxtfe'laio ThG thoory in Appendix Bo mant of ao'vonty-tvyo trials » sr.-snt ohoasn hy shcivs t]int the th^a vras derived froia an expari = It ccrrectiy predicts the incre~ subjcjot in sixty-on© trials, llie protocol suhjact was oonfuGod on tho first four trials and that ho explicitly raTiced his process on tho fifth trislo iJisT counting those trials means tho theory predicts 61 out of 68 trials o the theory This ie not particularly?^ surprising. hov/GTer, since vi&a doYelopsd from the dsta^ It should bo pointed out, too, that this record of prediction is with rospoct to the incremsnt cacaen and that tho 5ir.'!4.=HJ£Ll2. past perforinanoe v/as the basis of analysis at each trial. -39" Cl&Grly tiio theory could ba aIiov;a^ to opsrat© on its own experience but this seems to be an miraas enable procQcliiro on The first is illustrataa by two comitao subject's behavior on the first four trieis, about lad rtC<.83 ond to a confusion later (Stop 5) recogniaad xvas modification rulOo 3:j:plicitly ch&BS^a tiiia m.le. ii3 soin?. On Yet oa trial 5 clogrG© on all subaequent trials, bia prior estiiat vmicli he peviorica -^yjsis not that ;vhic;h Thus one rulSc to use vuiat oa an inappropricts goal by hin to trial 5 hini t3iQ ho nad on ths first four trials, nG'j.lO. iiava ar-gu-Tisnt hec t\ndep a more appropriate for txsing the subject 'a past behav- ior for predicting his nsxt goal is thot thi,? is tho informa- tioa ha used in his seal modificatioa ru-lea, To ignore this fact saesia to reqiAire tho theory to acooapliah soiaathing 'i.vhioh it should not b« osipoGtoa to dOo The seconol point vath r-Qspeot to tho eppropriats basis fop o.Ts-luatirxg tho thcaory's predicti-<rQ poY;er ia relatsa/Gut not iaontical, to tho firct. cess ir> one of goal rr.odif ication, closely Tho siibjacfa pro- not goal cyaaticno It seems reaaonable tharofos-e, to tost the theory vdth rospoct to tho bohs/ulcr it attejapts to oi^iplaiaj ioe., tho incremental ruocUfication rather than the lonfj run conoaquencos of that beha'irioro Tnesa questions on tosting raise quits naturally the probleifi of euplricsl validctlono source of clata for those toats? What should bs the Obviously it can*t ba tha identical oispor-iutsntal reoultG froui which ths theory v:as derived -40- booause ths theory This probloai is even moro porplsxing process, dascri^e only a part of the goal formation aUegoa to individual goals i,Q,, the effect of ijsrf oraanco on „ It the possible affects makes no pretonse of being aolo to describe ivhich are logically related of infoPEiation or ^sriabla classss to tha class in question, as a result even thougli it would explain a stream of be desirable for cxampla, to attempt to does not seem industrial budget and performance data this that budgets are not appropriat3. It ^vouid he hard to imagine against othor variables than past performance 8 f.motion of th9 budget. might sug» The theory dovelopod here although it problem, «ould have no gast an approach for ths study of thia variables o Thus.. vmj of Q3:.plicitly dealing with these other shed little light Q^en though deal rabl 9, such a tost could on tha QiapiriCDl validity of this theory to SincQ the ezpsrimental situation was designed that it night bo control for OAtraneous influences, it seems validation, Differont the boat souros of data for eiaplrical ponorate data and parainetors and suboaats oould bo used to protocols. '.vould If the theory could explain seem to offer this data this e-BipiriCBl support for the thoory. ^ Test of this form are eurrentiy plamiedo situation A test outside tho particular experimental the'Uovel of might b3 devised by using data published in Despite the fact that trial 'o:/ trial aspirotlon^^literature„ data ia not available, suciunary statistics on perforiaonce of statistics in the tha theory could be compared to similar '41- This test, while not as strong as a trial by literature o trial test v.'ouicl if succesarully passed offer soaaQ limited etnpiriCQl indication cf gencralitye SuTsmery The basic objeotive of this study is to that ^shoY^ relatively simple mechanisms operating on a rocord of past perfor?:enoQ can yield paraiuQters of the decision process v/hich have heen called goalso An 02tpsriraental situation hac boen devised to yeild beha-pior- under conditions whera parformance and goals are defined on g single -variablo class to avoid tho ccsplexity of hierarchiCGl effects on tho a^al fomation process in the. o a nuiabsr baaie expsrimant and been mada on their behavior o subjects differ in a fe-,^ of subjects have soa-a participated general observations have In particular it sesas that tho basic oonconts \vhich they ue& in interpreting tho feedback they receive o Thene diffaroncesJ in concept account for basic diffarences in the vmy they struct a value rscalQ on th« aluornstiva goal to them. An a:^;treKialy vali!.es con-- available tentative theory of the process of goal raodificatioa has heen eonstructodo this theory across subjects an:l ah extension of situations is expected to desonatrate closrly these conusptual difforoncas among the stibjectso Tosts of this theory are suggested in variations thoso of tha basic oxpsrinent with different subjects than used in the ezperinsnts to generate behavior for theory oondata structioru Further tests of the theory against published on level of aflpiration ozperiraonts sre albo sue.eestedo =42^ variations in the An UB-dersteaclins of the rorm aud proaeases in sspocted to yield insight goal moaii'lcation into a nuiaber of manesQmQiit problems „ For example ^ an thosa changea understanding or the problems of Implemeatins suggested by nomatiTs In rfl.8U0g9rial decision prooQSses modificatioHo may result from a theory of goal oconomc thoory managerial decision may.ing Data on the uaoconoaic variation in thoory of the phsnomanon may also be explained. Similarly a literature under variously reported in the psychological else result, xerhaps the narca'^leval of aspiration" may however, u-ill be mora important than any of these resulte, modification will the fact that a thoory of elrnple goal ths basic problem open the v/ay for a broader attack on allocation in individuals of underatandins intolligsnca end organisetionso APPSKDXX A Instruct. 5.02XS Tha experiraant in which you era about to partici^ pato is a part of a reaoarch project designed to invostigats certain fundaiacatal prooesses ox* huaan docision makingo During tho courr>9 of this expcria.ant you hoTS the opport'-juity to uake a numbsr of docisionso Thase dooisions >vili, ty a pi-ocosa dsscri'&ao. belovi;, detarslne a iceesura of your perThis msasurs vsij.l be oppressed forraance in the exparisientc can think of those cs your earncsnts you dollars ana in srA ings in the osperiiaant although ixo payoff will be n-ado. You shoulci ettozapt to vinke ;^'our deciaions in such a way as to males your earnings ati largo as possible On the fox-Es proYidsd you can seo that the azperiment consists of a n-ojabor of trials end a decision by you on each trial c Tov>r decision v.'iil co-nsist of choosing a r.v.iaber to writs opposite each trial nuciber. The nuDibers you choose can be thought of as dollara end cents can bs positiva or negative 3 and OD.n be of any sise you chooae„ Thore is absolutely no iirilt on the nuraber you chcoae on oach -trial QXCQpt that it be in the dollar and c«nt formate. . ji.ftar yoti have chosen a nuBibar for a trial I -.'vill corapare that .number with the nr?ixb&r corresponding to that trial en a long list (5000) nu-ibera vihioh 1 hava prepared., Koto that tha nuiriber on my list changes at each trial v^'hsthsr yours does or not. If, \vhen 1 coarpare your number ^vith mine on any triel, 1 find ray nuiubar 1b highar than yours 1 vrill say you win^ In this 6'9'sn'u yo'i will circle your number and note bssida tho curaulati">re nu;K;53r of wins up to that trial,, Tou can think of ji'oursalf as '\vinning tho money iraplied by these cirolsd nui&bers.j If on the other hand, the miinbar on ay list corresponding to tha triesl is smaller than the number yo\A have written on your list, I vyill say you Icae ana you can proceed .- ~. to the nosit trials You Vv'ili contijrao to play >intil you have vjon 50 nuiaber of trials tJiis will taks ivill vary depending on ho'.7 frequently you vvlio Trie iDiniiaum number of trials of course, is 50 but thers is no ruji-e as to the mesirauja mribsr, Tou Slay take as many trials to van 50 tiiaas as you like or find necensary. Do not feel constrainsd one r;ay or tha. cthor by the fact thfit tha prepared forta has 100 trials on it. You may use more or loss trials if you likSo tim-aSc "^^'he In ordttr to livait thfi number of trials you mey find it desirable to play, ho'.ve-ser, a fcse is iffiposedwhich you must pey per trial v/nethar you v.dn or losa« iJinca the fee will vary from ftxporiinant to exporimant you 'ivill b^^ told the size of the f OB per tjilal bafore you bogin to play.- For Q7,auiple, suppose you pleyed as fol3.ov;s and tho trial fae xvero i> oiO par trial., Trial_ffQ;i 1 2 3 iJac_isioi\ 15H?£li'2°B^_£E>,^^E2B£2, o75 o60 o50 «50 4 Lose Lo2G Loss Win Your earnings through trial 4 would bo; Svm of circled \'alues o40 ~ .Four trials at $OolO per trial .,50 = oiO •- Earnings through trial 4 Are there any questiona about procoaurs? on my lint IIoi^; juat a v^ord eboat the nuiabers These nujnbers v/are seloctad carefully but then arrangad in a r8-i)4c3 sQCAxenoa such that e.ny numher ean aoIIow any nunher as ono '"goss dovvZi the llsto Tou will ba given no information about the set of nujnbers from which this list ;vas solectod^ Since this esporlniQnt is deaignsd to reveal cortain aspects of your dscisicn process, it would ho helpful if you can desoribQ tho process by v^hioh you select the nurabars you write on your list na you go alongo If you fcrgat to do this, you may be reminded by tho sxperiiksntorc Since most subjects vdll; like yourself coma from tho school, you srs reuiestod not to di/jcuss your QxperioncQ hare outsido this roos-a because prior knov/iedge would, no doubt infliiencQ future subjeot.3 and nullify their velua to this studyo j v/ithirA J APPMDIa b -"-ppoiKJix Bo 1-aga 1 <, Tha follo'!7ing dlecrimlnQtion net repreaenta a tentative thecj'y of the gyal modification pyocssSo It V7a3 derived from the behavior and protocol of a siriglo aubjocto To disGo%'-ar the predicted modification in goal value 3 epply this net to a record of prior performance o discover the predicted goai value on a given trial apply the predicted modlfiQaticn to the precoeding goal value Ths nuiuber pracaoding each question and answer indicates tha ausbQi? of levols that quootion. or ans-wer is from tho too of tb3 net.. To '^ o lagQ W ^-Doendix B< win at latQati goal value? 1 Yea 2 Prior trials this lei?9l? 3 Tos 4 Prior trials at higlior levols? 5 TeB 6 irior vlna at higher lovels? 7 ¥06 top? 8 l5 this level more than ens -rrom . 9 To3 10 Go UP 9 on.B step. Eo 10 lost more than higher levels? lia%'3 -won at 11 Tea 12 Have ro?i' last 4 times in this lavai? \7on 13 Tea i"^ ££ 2£ 15.2. l£:i^'' 13 Ko 11 Ko 12 Go 7 lip one atepo Ko 8 IlavQ played more than one lorel higher? 9 Tee 10 UavG loot at top more than 4 times? 11 Yes 1*'^ 22 HS £B® SH1° page 3< Ap -ondix Bo II No 12 Go to \02, i22Si.° 9 Ho 10 Is next level ons or mors steps higher? 11 Tqs 12 Go to ne5_ JrlL^^Si^lSlS. "XeWl', 11 No 12 No Chang a la S£q1. 5 HO 6 BQva 7 won bofora at this level? ^ Tos 8 la this trial oa an ©?en dollar? 9 7 No Yes HMSH." J-0 2E H£ 3-0 S£ H£ 2»1 Sl22' !^E£ . previous S Is thlG trial change from trial? 9 Yes 9 Ho dollar? 10 Is this trial on an even 11 Tea 12 £0 ?^ !^H£ £5-^.' 11 No IS Go U£ ona atoj^o Page 4o Appendix Bo 3 No levels? 4 Prior trials at higher 5 Tea levels? 6 Prior vjins at higher - 7 Yes 7 No froa top? 8 la this levol less than one 9 Yes 10 Go to topo 9 Ho 10 Have won bsfors at this lavsl? 11 Yee 12 Go up half cit.££o 11 No IE No ohajiga in^ £SSi.- 5 Ko S 1 Go uo one stopc No E .Vrior txlals tnie level? 3 Yes loYsl? 4 Eavo ^on before at this 5 Yea levels? 6 Prior t:?ielB at higlior 7 Yas level? 8 This trie! decrease frosi last 9 Tqs Page 5o Appsnclis Be 10 Was previous trial at top? 11 Tea 12 Go do^ on© £te£o 11 Ho No 9 10 Have ^?on on edthor of last two trials? 11 Tos 12 lio cliaQgkO in £°2i' 11 Ko 12 Go 7 dOY.-n one ISZSi." Ho 8 Two or IQBD trials at this level? 9 Yoe 10 Go doua ^18 ^^22." 9 No 10 5 llq c>iaBge in goal o i^io lantji^^al a w3^= 5 Go to next 10^^22 iSSSl sSSro 3 Mo 4 Is this trisi an oven stop? 5 Yes 6 Havo played etep bQlo%7? '/ Tos ? Ko 8 Go fiovm (TTia stOBo iagQ Append is Bo 6, 5 Ko 6 von Hrvq played tilgbar levels and not 7 Tas 8 Go do^n 7 M^f ste£o Ko 8 lio chs^a in £calo ^ Bibliosxaphy lo Moti'ya= Aao.areoM, HoH. an<X Bx-aridt, Ho Fo, "A gt'Udy of «*i=C. 2. "Gonslstenoj- en^. Opt inality in Managerial K. Ho, Bo^^i^an, 1935 o 4c EmwQr. 5o Cnat)S0Si, 'To Vol 54, D<, So. Goouno-v, end VJo 19i>9; Jo J. ?ollriiian, and Aus'bln, Go Ad. 'A Jo, "A Social DQtQxmX'mnt o? ppo 325"SSSo ^i'o Sr-oo3?v of C.oaa-u,:^-3:e D^Hsna: A On "?,. jlo G??itlGel'ApT3y:7^isal, PrsuSia© Kali, i'oFbiicoming, i9o^o ni n-^'icpop fi t,it-J.%,«.-L-*- FlTKi/prexitlc© Hall, Bo Tismbo, To, ^^Bs:? ?^£SSSS£BS; '^oX forlilaoovdiZLS IS^^c /bgeral BFna::nir3ahas i^oblems^ PHIS^£i£E:£S£ IS 1951, ppo l-^^o J lastittitQ of "icaclxiiolosy, 1959. FGOtora J^^Siii. iilo < Le^ei of /.Bpxration", joHSSl- §22iSi ES.;i2li2i£^» ^*^^ ^"^ ^ 194:S, #rr^ ppo J.o^-tsOOo Jo Do. '^Inclivriaual Di'i??or©nco8 in Q^/ffteln Aapae-os of Level or As-oire.tion=%, .^aicrU?aa douxD^ _^ P£?X5i}2l2SI^ n^-\nli, Yolo 12 ,/^f:e0o;;iJ2s 47., I9Sr?j '015 o 110-1860 "\ "Soms PsyehologicaX J.^Qtsnainants of 19S5j ppc 285~E9i3< i--ev.3l oa c page. 2. Bibliography the Level of lorformance 5E'^5~ta0k on the Level of Aspirotion In anotner, l^*J''. Journsl of i;:;perXmantal Ps22]lHi2IZ' ^°-''" •'•^' pp 159-17X0 tTThe IndTluenco of 15 " 14 Properties «A Cogiparlaon between Certain of h&^^r^' Aspiration and Random Gues3ing% Journal ps^holo^',.^, X^olo 'i, 1937, pp, 4S-&2, ' of 15^ «Rocant Studies of Level of ^spiration", pgy^cFoIogieaJ-. Bulletin, 1941, ppo £18-225 16 o University Friefean, M,. Sssevj^ in l-oa3.tive Sconoalcs, of Chicaso i'^ssT, 19530 _, Gardner, J. W„ , "L^^vel of ABpiration i^/';!!Pf.;^^J'^.^..n journal 01, lt;^±iS£BH£ Pr©Qrran.<7,ed Sequence of Scoros% 601»62lo ppo Paychology, Vol 25, 1939, 17. 18. • 19c Variables "The Relation of Cartain irarsonality tTTS'^^ of Aspiration/* i^^rn^ Sl £^S^^21S^» """^ ^' 1940, ppo 1S1=-E06. , Goiild. R„, , EO. 21. ABpira-factors Underlying feprossod_L«vol of ''Sotae Sociolof-ical Doterminatos of Goal Strivin68% "The R^lf if ^^^'^P. '^^..^t!:;?-,/ Gould, R, and Kapion, lU Factors';, lournal o.^ ^^o^^^ Porcsonaiity and Aspiration 31-iDo PGi^^cnoXogx, Yol, 11, 1940, pp. , op pPo 422-4vd. of feoOTOJjiantea i:£X2^oloiit, Vol S7, 1940, E3o neatbsra, L. B., "Factors rroducins ?^f f^^J^J, J^Sv^'^Yol 50, 1942;. pp 392"'4b6- 24. 25o E60 Hor^pla, C„ G. B.nd Cppenheiia, r 0., "The Logic of Explana^ Henderson, J. M, and v^andt, R. Bo, MiSZ£S£2S£Bii Bl£££X^ JucGrav7 Bill, 19580 feplicit Goal. Hsrt.man, Ko and Fostinger, L., "Shifts in o^ ^^n^^ in a Le-v-Ql of Aspiration xixperiraent-. Journal 4o9==^o2„ m^M-, i-sychology.. Vole 27, 1940, pp= Bil^liograpb.y ForsGlirmg, Vol., 44, Tflv'^n Oft ^ - L rf'ml)0. J„ Mo Hunt, .19i^0, pp- i-Sr-ic T.. li'estinger, Lo^ and -Soars, t. isditor, Roaald i^raea, vo.uo ^ x, -ya^o £9. Wiley, 1959„ Marcb, Jo 0. and SimoB, Ho A., Or£:anisatic^, 30. Newell, A„ and Siaon, H. A.. "A ^^^^J;? -J^^Jff^f ^^of 8 July 1S30, pp. 10-17 o- „, ana . ""S^® Siaixlation of Humn Thouehfy of ?6;i2:X. UnivcrBity • cp:?iM^^^iirir^il2i2S^ ppo lo'd"l-f9o Vittsburi.n'' 1TQ33, 3So IS&is a }.Qthod oT Rottor, J. Bo, "L073i of Aspiration oa vol. 4., Studying Personality 1% i:3voholosioai ReHiS, 1942. ppo 405=474 "I.evQl of c-c: Aspiration so a J-ethod of^ Pa^cholo^X, Vol SI, 1942, pp. 4iS~4a<.o 34. Russell, Bo. A Hisjiori of lostern PMiosoHlI. Schu3t.©i% 194&0 -55 Siocral ^R ^ moa ^^^^^ &; DQcisioa Laklag," So, "Lavel of Aspiration and "Thoories of Dcalsioa taking in iSconomics American iEoon^Sio Hevi^Z^ ^inia EaSakoral Saienoea% 25b-E8ii = Volo XLIX, Juna 19b9, ppo Sliaon ^- /o. '' Date Due AUQ27 '83 0CT2I79J9 Lib-26-67 LiBfiARIES MIT LlbHAKlIL3 3 c,Q6D ,,,, .•.|i| I I QD3 602 670 MIT LIBRARIES 3 TDfiD DD3 fi OE fi5^ o^~ ^"^ DUPL MIT LIBRARIES (9^-<^^ DD3 a33 TDflO 3 flD MIT LIBRARIES OW-6 2 lllllill TD6D DD3 3 fi33 fi75 MIT LIBRARIES liiilll 3 TDflQ D D3 flOE ^^^'^^ 651 M!T LIBRARIES OC^^^ III 3 TDfiD 003 a bfi DIE