You are on page 1of 17
=: «=««__ 160 TRABASSO ET AL aaman fA Seibert, 5, Chses and cileoas Internal eramsation and ea I ee pan 1976, 8, 561-577. aay Ca ere clasinchaon protien formation procering trp Senior fionars Theas, Princeton University, 1976. - edamiaitaaal sag in RE ae dn ‘of the child's cognitive and linguistic development. rch Pa EA is et ns te Surat of fepenmentl Cd Pay. pine) Te cs conception of mtr. ew Yor: number. New York Mame res 19 Hard Si pryehaogeal tui, [New York: Random House, 1967, = ae Bathe neat srucore of Prep! an somone Moore (Ed), Cogurrire development and the ecquisinon tee te deme Press, 1973. * eee semen Be Eegkston Ve thy & Seott JL, Nani : ; colt Jets Number development young hile ‘Cognitive Psychology, 1974, 6, 357-379. f Saat Se ecpenmantl exertion of the Pigeon cls incunon tas. (Techov Report 18. The acquisition of lingurstic structure.) Unpublishe Seca ommad epi tea ‘nptished monary, Ueesty Sn andcnon, R:C, Herchletsemanuc cxpointion osm Yeu ol est omelet rene cata 1975, 19, 544-553. eee Cee Stating Steps and semanive nays sapped oes ictus. . “tive Psychology, 1976, 8. 64 85. a an inh acolo eel ftom of css non reason ‘ment, 1974, 45, 224-227 ee olf Responses ees eh cles hon questions or etal and pctoraly resent ‘tems, Child Development, 1968, 39, 449-465. Goal Formation, Planning, and Learning by Pre-School Problem Solvers or: “My Socks are in the Dryer” David Klar ‘carnegie Melon Unversity INTRODUCTION st 2 report ona propst aed at understanding 20 ultunately umprow Tee poblersoiing biives of young cildren, Tes ‘path practical and oo eepea rounds for such an effort The practical CO ‘eames from the Leo er prablen-ung ais are eptety aun arly school coer: or example, inthe puzzis and gies wee teach and test funda si of reading and artbmeic. Not only are rudineviy problem-solving ee ced av the ely cae, but aso advanced general problent ssh aks are an explicit goat of subsequent Insite ‘We have al heard th aaa er rang mn anahcmatis euding (OF YOU favo subyee) enhances ea iiy of students to nk lopaly. Ws semen CA then, that For the a and explicit empha on problemsong 2015. 64 racly sath ety to young clden. One fongranee goal oe project, then, 1S tase dee rien on general probtensolvng methods sett Pruatanterest much » stay derives (fom ¢ ook te other side or eer ten Ten about problem soln, even wivens 000 anstruc of cn canon and cua observation indicat hal 6 ATC approach, sol ages ey seq 2 ange of problr-sotwng sii that are typically Characterized simply 28 "common sense.” ee ar example of wat T mean by commen ans feao2INe ma ‘young child. Consider the following seenano: see cid and father in yard Cait’ playmate appears PS sen ealdy, would you unlock the basement door? Davy: Why” CChikt Cause # want to ride my DIKE vt vez KLAHR Daday: Your bike 1s the garane ‘Child: But my socks are i the dryer What kind of weird child as this? What could possibly expiain such an ex change? Let me propose & hypothetical sequence of the chi's mental actiity Top goal: ride bike ‘Constraint: shoes or sneakers on Facts eet are bare ‘Subgoal 1: get shod. Fact: sneakers in yard Fact: sneakers hurt on bare fet ‘Subgoal 2 protect Feet (gt socks) Fact, sock drawer was empty this mornin, Inference: socks sill an dryer Subgoal 3: get to dryer act: dryer in basement ‘Sungool 4: enter basement Pacts long route through Rouse, short route through yard entrance Fact: yard entrance always locked, SSubgoa! 5: unlock yard entrance Fact, Daddies have al the keys to everything ‘Sungoal : ask daddy ‘the example 1s teal (in fact sis from my own experience) and should Be puible o everyone sho has spent twne around young children, On the het wand. the analysts af the example x less convineing, based as i 8 on & host OF assumptions, Some wf these assumptions are easly testable, We could detesmaine ‘Ghethor the child knows constrants, such as the one about riding bikes ony When shod. Similarly, we could asess the chil’s knowledge of facts about dryer IBeatton. shortest route to the batement, and 30 on. Somewhat more dificult, tbat stil reasonable, would be the yob of finding out what sorts of inferences the child was eapabie of making about her day-to-day environment, such asthe one Spit were the socks mugit be, given that they were not in the drawer, How: Seer, the dominant feeture of the hypothesized thought sequence is not any one Sf these features in wolation. Rather, i their organization into a systematic Sreans-ends chain, Thus, I am suggesting that by the time the child 1s old hough to exhibit the sort of Behavior jst described, she has already secured same general problem-solving proceses. These enable her to funtion effectively seiThat is, to achieve desied goals ~ by noticing relevant features of the environ ‘Rent and organizing a wide range of facts, constraints, and simple wnferences 1m some systematic manner. ree Mt stands, such 2 suggestion 1s uncemarkable. The meresting questions concern the detailed nature of such processes, their generality, and thet develop soewtl eeurae, Parapiasing, Newell & Simon (1972, p. 663), the question #8 1 +. cHILDREN'S PROBLEM SOLVING 183 vere ean he il se ok anime OT sat ag in wt em HY cach eit oti, pal srt 2HE Yptwely prec fasion sae me wh pec of een ‘rom Fae rane om caress pn on ay PT he te eo ste tm an etlegel peroee nd eee the occ of i lo me Be cas ean ha hyena Kes ca sear ea a a mean Teas cy =e elo He cat we a Sem 2 eg ee aa ‘oN a at ne I a aa tie sarc fe ena enone fom eu ars, and playgrounds : seam nt phase of ths prope! fs Toca tne performance of huey betreon the ages of 3 to 5 Yearson a ANY of well defined take The saree an the sees, and the only one will ePoHt OF ete, 1s the Tower of Hanoi “THE TOWER OF HANO! fe pegs and a sel of re standard vernon oft tak conse of #5 of ES 8 ‘Take of decreasing size The disks sit italy om fone af the pegs, and the goal ere ane enue disk configuration 10 200410) Fy gubyect 10 10 COO" x te Only one dik ean be moved at a tue, 2080 ‘ant can a farger disk smnts: Only Cer sk on any pen pas A standard thieedisk problem 1S showin in Fig. 7.1. or aave this problem you might reason 2 FOUOWS su te stack vp fom the bottom, which means BA TEL ae! Tit have to move 2 to B. But prtdak3 from A to C, but Bossa the wa 50 Tucan ws af tn tat evel soe oe AE BG methods of expe “lt at en ene 8 SP 1937, Eom & men eel) 2 sun, 962; Homann, 1968: 172 tives, 37T,PeLeom, oe arigte sino, (995: 5y0™, 1970 184 KLAHE z = t Problem: Move oll the disks om pea #19 pe0C. £16,741 Thweedisk Toner of Hanot problem set want to move 2 to B, | must first get ( out of the way, 90 my first weve ait be | to C, Now tet me reconsider the new configuration: To get see still have to move 2 to 8, which I can now do. Now toget 3 to C 7 must reaove I from C,s0 1 will put it on Band at last | can move 3 to And 30 08 ‘Aithough, a: wil become evident, there are othet ways to solve the problem, the rample shows that even this simple version of the puzde can tax one’s sbi Te codinate sequential reasoning, perceptual discamiation, quantitative order tng, and shorter memory processes. The task snvolves a well-defined! iia ene, an unambyguous desired state, and avery Hinated set of rules about how to Ghange slates. The difficulty lies m oxganiming a sequence of rule apoeations (egal moves) that ultimately transform the nit physica! configuration into the sesired one children’s Version of the Puzzle: Monkey Cons For use with young children, we modified the tasie im three ways superficial appearance 1 changed its ile mauntainng te baste structore Matenal, We use aset of nested inverted cans as shown in Fig. 7.2. The cons are modified a0 that they ft very loosely on the pegs; when they ate stacked aipsit is impossible {o put a smaller ean on top ofa lager can. Even if the chi foagets the celauve size constamt, the materials provide an obvious physical con equence of attempted violations: Little cans simply fall off of beget cans: Furthermore, the materials ate intuitively mote “reasonable” in two regards inst unlike the standard problem in which small disks may obstruct larger ones. In these materials, bigger eans obstruct smaller cans, ether by siting st0P ‘hem aw by heing ona goal peg. Second, larger cans not only sit on Lop of Dat vie partly conan the saller cans. Fach cap 3 different color and makes # Sava ang hak wth each owe 2. cninpren'srnosremsocvine 18 « ananged for a onemove_ problem. I ne ane eamirtion = ante # ee E73 in atv to he eae coniuraton He ate atssinn ee DY eso te ads cs ration and the experimenter’s ‘cans an the inival wel > Son caret conti recent i can ami casted co CH A ee cnt, Th arte oas(E29OO ie cam rac MME OT Ts ee a pre hh epee sen eee externalization of Final Goal yummy from wee to Ue (pes to PPB) "re chi’ monkeys a6 ne as monkeys Formal State Properties sand al egal moves for these materi: Fe at space No configraion 1 repeated the 27 sates The cae alcated by eeled numbers, and the can thot moved inetd sates are diate Mae connecting ajcent sats, The 1 to problem the mt ome a path a tie of ated) C000 the slate space. Figure 7.3 shows all possible taal state 185 KLAN, £46.73 State space forall vince problems example, he mimimum ston path fo th ccm 1 forthe problem tia sath ll hes Cronous oat se sce ng fatness oe ig eon a > 2 ean pop (tate 3), fallone by 2 ove of cm 3 peg B (state 4), and so on. a aby aaa rc A a a Se ing an rbary ail and ial ate, However, thee at saat states for which the runumum path requies more than seven at "ha a a and se! mn rh A ta ce pblens“avenigroblns, nth *sandard” eet rove problems “towerending” problems” As we wi ane alo 1" problems.? As we will se, they have somewhat Trower of Hawo! bus should pote tat the “monster aioe at stan of Sion a Het So fal eer ti teats wh ‘nl having all he obecs-0n thE tere re wads toa shorter oiuon path then when one of ri a indy (peemavutely) the rt focation eg..13 {6 301 1710 6) 2. qmmbonen'sProstem sowvinc 167 general Procedure se gener procedure desenel 19 PE ‘the upper limit (measured by the Te the mu oon Path) OF sildren’s ability to sve this prOI, lead 8 ncoduced to Ue mater the rues, and the cover story and Frnt wh # oneove problem cae Fig, 72), then a twormove problems snd 02 SOLUTION STRATEGIES we can expet chien to vary wide 18 atotiy to solve these problems We i if we ook aed to the mes BAT “Jesenption of the children's per Inte, we can se om Table 74 thal est subject could reliably solve formar problems, wheteas the poorest SOP {oul do ew better than 199: seer rable, To satrpretsueh ena, redo propose hypotheses about mors ge processes that enable» TESTINXY) Ti, list <= Seerom() F2. Llist <= SeetolY) 73. it flist = nil & Uiet « nil, then culprit <- nit Ta: else culprit < minis. it): exit F1G.74 SOLVE: A set of rus and Lets forthe “sophisticated wer optus srtcey Figuie 74 shows 4 concise seniformal desenption of ths stategy.? Ie addition to the 2 numbered steps yust described, there are four “test” steps that describe the details of the “determine” an step 4, These covrespond 0 2 senes of perceptual tests used to determine whether there 2 culprit Blocking the gurrent goat and, if there t= mote than one, winch one should be dealt with fiat TI notes aay cans currently above the sem whose move « being consi dred (ie..an the “front” peg). T2 notes any cans on the curent “to” peg drat sre Ineger than a, 73 tests for whether both of these ists are empty (ail) I they are then there 1s no culprit, TS chooses as culprit the smallest of the bbstructors on the combined £. ("from") and t. (*to") ists. Figure 7.5 shows the first several steps that SOLVE would take when pre sented wath the sovensmove Mlat-ending problem starting with state 13 (2/1/3) tt ending with state 3 (1/2/30. All the steps listed precede the first move, The Ta seven snoves are sown in Fig. 7.6 (im abbrevated foram). The figure shows the sons of goals and tests that precede cach move, For example, lings 9 t0 15 tn Fig, 7 show thatthe first goal (line 9) 18 to get from can { from pep B to og A (OL:IBA). The test detects that can 2 blockang that goal, ThereFore, {ine LI a new goal (G2:2AC) is generated, and so on. Notce thatthe fll Ii STemodetewFigs. 7 sad 7.8 are written a2 sequence of FORTRAN or ALGOL HE stepe th nnch tere are seca bouts that are undeine but whose funtons ar a Sere ont These models ae not writes 35 production systems simpy because the trom commmealyrequued by ch a rapresenttion unwarranted bythe lve of 88- vreweare wang ete (et Rlahe & Sieger, 1978; Sion, 1975) “Sinema (2/0/3) 8 sed te and pox € by 3. As anther example, sa ne that pey A wsoupad Oy eam 2, po HY L arn tig 7 3 would be described as 31 2. CHILOREN'SPROBLEM SOLVING 189 range C ine & cae C. initial state: 2/1/3 {State 13) & tint star 1/2/9 Glee 3) gi: What's wrong? 1 nat on As 2 not on B SF. What's the smallest misplaced cent | $3: Get 1 from B to A 54: Can 1BA be done? Ti: nothing on top of 1 To: something in the way of Lon A: 2 Ta: 2 is the culprit. TE BE ar eut of way of IBA: Got 2 from A to © $4; Can 2AC be donot Ti: nothing on top of 2 Ta, something in the way of 2 on C3 Ta Bis culprit TE. Gai out of the way of 2AC: Gat 9 trom © to 8 daa et constrained objec.” 1 tear get 0 where OPE, the rest will Be Tha mages C976) sey ofthis pe, ren Were Te sit na ‘Tae 0 ete eed ee on ne PONE OF et mn es I ee end oy ae een staat 8 i tytn ae te enn ae eal meta apcinn of enon a He ene the ln, he pee onto Te He 90 KLAN ca sane re ein tin Soar ras Mowe cont ° ra 09 G1: 1BA oeec $0 Tet rar (exo. q ‘crane 2 See oe 2 on ee “ Fee tom i, xb. ® xe anv. 1 1 arto, Xe To rar (02h: sma ® on Bahama. 2 » oso i.e oe ze -rua #2 Fes tom oo, 24 Ga: 38 " : Es ft hom i = tut elon 7 as we ora ‘ham ot ce snare wa wena rs 3 Fone tom 2 ttt clo = ere % See om lab * yea avert 02 2 os-z08 Sa Ta tr ii, = 0 fa, cont 40 G7:3Ac ven saat OF Sm ott tier ase 12 ou £16.78 Fulltaceof SOLVE on 13-3 Navending problem. con nt ye on ts goa ep ected 5th one Ba ec conen pecans tc on gels chosen the eu alton pe, Ts fa, bles have been characterized by their minimam mn feng. Has also instructive to consider the type of moves that need 10 bbe made. There are the cae. Thee he ve YE ‘dstngusshed by the immediate reason for 1D mmaves: Move wean directly 0 fina goatena Me ‘ean fram a peg in order 10 get iv one hencath tt ae ae 1. cHILOREN'S PROBLEM SOLVING 197 sr moves: Move a can from a peg inorder (0 facilitate monng diferent con to that PER. able 71 sows the sequence 19 ge) moves occur along, increasingly Tan cowering sition pats? THe rrtest problems consist tely lone ret vo te goat conte Problems of lengths 3 start with ENE oF of the gest 20 rom on 10 0 Wodalesized can 19 ordes to miO¥e tere mide can} on the next move, OMT rotiems start wih the direct 3 re te smalls. cn to the gel EBs [Notice that ths classifica- rm unoves accocding 10 why they ate UKED ‘smplies some sort of goal 706 Wry on te part ofthe problem sve ee fortis csifcauon m eral BY CH examination of Fig. 7.7 arrne ae Tor the promote feaon fo" a7 TE te answer 1s obtained BY A weag for the wuesintesuperaoal thal he cccomplishment of the current wae il allow. Working from the BUG he top of Fig. 7.7, we see that apd Dd have no supergoats other than Oe replat solve” goal. FI moves 30) ct that was diseowered @ be ona OT" TE Hor the preceding goal (2CA) et athe deck mone. T mmoes can 3 INE ths Lime because it a8 08 ape of the supergoat, F2 and F3 bal wre moving cans from the from el the especie soperpels ASP dsiicaion for AMatnding problems ern the gost clue EAE) Notice that for a pat feng 3 Sho Tour mnoves, the Haending problems NOE A eiferent sequence of MOVE pete rom those of the tower ening oben Alternative Models she SOLVE model wt Fig, 74s wty powesil wll generate the mantras pat solution for ay thes, ety Fing robles and for any Maik, owes vet epeaben, Ths, can be wewed 8 9 posible “eogave obyel¥™ {Greene 1976), that, a8 the vitae “if rauming someone co perform cerery on this task, What about othe, Tmnctionally equivalent strategies FoF tt performance? In ths section we SMTA three quite diferent strate Fer dtornes detail by Sion (1975) fo perfect performance on the TOWEE wes dese eur that none of them ae 3 HWE be acquried by Our ayets asthe spstcated perceptual S13) epiozented by SOLVE, Then, sabes aon, we describe a sere of nese ‘powerful partial models ‘fat eulmmate an the SOLVE model “The goa recursion strategy solues the edb problem by eaten parts: (1) removal of the t= di pyramia Me) moving sk from teal £0 the Gos! Recursion. recursively decompose i Thom the inital peg to the other PEBs Tre wamibers fll the Mowe YEE a pam toger problems ae presented: THOS) {hind Fspove, and 30 forth diene te onder of cue 0 RL TS OF Mie second Denon, £38 1 192 yAaLe 71 Move Type Sequence § a ar add ath Path Lente Fina Iieat q 3 + cuiLoneN's PROSLEMSOLYING 199 ga 18 PY SOLVE ate Tomsr-anci 0 Oe inate atl Fle 320-4 cont, TYP" 6 o —rav- 3 ro oo GEA 98 See om (9:2), 10 (ni sales 2 a u (92: 280 2 Fee ror (2,10 (ean. 8 "23:80 “ C230 sn ni cust, 30A NT F S Ye Gh TOA 10 Soe tren 2 [satan 2, 2 e (oz: 280 * 2:20 tonic, 305 MRT 2» Gu tga 2 Fae trom i,t (3; eee B ‘98: 386 Py oe Mam il oiler Es anc 8203 2s PN 2 Fe tom nt carn wa v-nz FL 2 no 5:20 38 Fe: rom (3c (ni, ter 2 (96:38 = GE SCE ap w.cceren 000 WE 35 5: 208 Be Sea: from il to fis etait aca 2viat= 01 3a G7:308 35 Sen: fom ft 0 fi cent oa 22-1 goat 9.27 Fuliace of SOLVE 8 15-1 vorerening BORER sat postion and (3) morn thet disk pYTAMIS from the other peg to the that pee Steps 1 and 3 a accomalshed Tecursvely applying the S8me searcea-Two considerations make a unity Wal Ot subjects acquired this rey pt hey were given only mite ENDO spycamids” thal seem water mace the approach. The only exaMPK ofS they encounter neces 1 Mowe, vowerendng problems Second #5 SIT (4975) notes, 10 3 the a atgy, the fl goa tack must be eae ‘memory. Its both aera pad sient for sation: “AL cack ee problem solver can neces ext witout any seerenee ro he cu distribution of the 194 KLAHR save. FE Swecen 26 em, te ene heee ie SOWEAE, 6 EL Ftd termes (5) Mn, then re Eb pi feeue eles $5 fewgen = oor BE Gio eRe ge Whe Reiecand est ‘he WovElodp caveteespih mph ft rest ae 15. pel ett soe 9) Fd arene (6-3. ana, ten dom Sb ne ae outed 3 Remgeot= Tetrsae = int hen MOE See Boweiatan cuvette, a rest Siaen 12 tats sene 13 ik a ce tet e'sont apt oa F1G.78 Part strates: SOLVE.2, SOLVE, SOLVES eats me pn a tan erm tf ierarchy m memory, he can calculate what needs to be ds ot the puzzle and without holding a ety th pe nwa ee guna ne memory demands of pees chien, Notice that the memory demands of th noe Jemands ofthe sophststd percept sae si paar nas eae ony geoph the 0 make the move directly associated with mae ly assoctated with it oF to generate its unmediate sub fie strategy, it seme unlikely 10 be acquired by young ae ye gal Als amore 2a imagined), the entire procedure 1S testarted and the differences between the actual (or wnagined) new configuration and the goa sol or ape guration and the goal (which 1s physically present) 72 + crivonen'sragucem sorvints 19% ove. The move-patirn strategy valves ree Semple rules that allow one one ta ste any tower ening problem.” (1 Y8 really want to ampress our ends, you shoul Tern ths and then SPP) ‘sovensdisk problem, ot owes wthot 3 ith!) This stacey, Hike Moe 1 for the balance scale {Seger 1976) 1 ens to remember and exci, Wt 2 very haed to duce, (Se ul, Hw quite umpaustle that our eifer ‘aoquired it ths study 1 eer fon of vote steatey 8 based oma memorize ‘of moves. To solve any apc omerendng problem, the subject OY OST through the move Ay ce he has memoraed for that problem. Ste} strategy could include ae Me ot generally by beng couched m ts of final, and other ‘i, whoch ave then wrstantated for ay DMN indisk problem. Another ee te aaegy ould const of» Tage ealecion OF ‘SR pars an which for ceription ofa specific cuenta par of cong OTe and R ss the $e opnmte nove in that tation (© “sate § V0 Se 8: move 3 to C"). sop age number of such specie uoctOnS an limites exposure ine ge co ros of tem, tis ges to Ue am ONS sequistion. Fre Ane ees teal not expla how subyets wna have Sn for example, prob- Prete and 3-1 amediatly sole 6-8 or 13-15 partial Strateaies ve have reeied as unlkly the nagoralternative tales Yt subjects mght Hea they become expert ths ask, Now Tel Nene characterize afferent levels of performance 1 term of weakened Yo ‘of the soph a aa ryeeptalsratgy (SOLVE) pst deseibed. Fiz 8 shows three such arya strategies, bn each of the models eps ave Peck tumbered to correspond para ee SOLVE (P74) Each model w named store 19 the length ao plem that recedes the fist ear she model wall sake Ca a sole up to wosmove problens. but I wil 08 all longer one eest BB. tn step 2 i determnes whic of the GIG involves the os eee nd it establishes (in $3) the goal of mown Tee directly to Mages pee Having set the goal, hen makes the om (S4) without any its goal Ee yal testing ar comparsns, When presented 8 anything other further Porerjern, SOLVE.? makes the wrong rove besa ‘nmediately ana ae ta move the smallest can nt yet on the gos! PEE, For anything other simon (1975) deerbes 95 Flom na steve mots new ie amet GH: ON TENA SE Gey saan oe eet die ttt expos: 3. et PRBS We heii source Pet, T iene he nent pe tere Hen he ol me odd the the ne a Owed om 8 tT 120 (0 an 00 srs fs the smal! ik aways mowed ms PORE de Tom's to 0 fo T to Sand sn fp. 2731 196 KLAHF Tale 72 Move Selected by Fart trates for Each Problem TYP ‘along Tower Ending Path oats 3207 =! Conf Tara Sater dtp IBC «SOLVED SOLVES SOLVES SOLVE Bt BA 3A sua 38a 2 DL RA 1A 2A 3 OR eB CA~ 38 38 a DL yn BA wa WA 5 oP mR wa Bac 3AC 6 Fy BA aact anc $F "pai. WAS aeavorance 300° 3A Ties SOF naman pt Anan D probiews such moves ate ‘lel. SOLVE.2 differs from SOLVE (the lnodel ur Fig. 78) an three respects: (1) Ht does wot test the feasibility of the vrowe it wants to emake, €2) the only goal st sets always includes the ultimate fal peg rather than any temporary. anieznally generated subgoal peg, and (3) ft never deternunes the smaller of (wo obstructors SOLVE will solve all problems up (0 five-nve problems. The steps ate sunilar wo those am the SOLVE model, with two unportant exceptions. hath suMfamed an 6, When a culprit 1s detected, S6 dues not establish new goal soa hen return to $4. Instead, it mediately moves the culprit, Furthermare, $86 does nol lave the concept of The target peg for the culprit s simply sh empty peg, Further differences hetween SOLVE 4 and SOLVE lieu the very aes tests tat ere used co determine whether a move can he made. These tests Terme only whether there 1 anything on top of the can to be moved. If there 1s nothing there, the move will be attempted (Le, there ss no culprit i the fists empty). SOLVES will solve problems up to louglh $ correctly and then will hegin 10 makes moves that are not on the miimum path. Unlike SOLVE, SOLVE.5 row uses the concept af “other” when determining where to move the culprit, fut like SOLVE — i makes the move dicectly, instead of generating a new foal, SOLVE.S also ticks the Tall testing capacity of SOLVE, In parteslr, SEEST 5 does not determine the smallest of the potential obstructors on both the Tist and the Fist. Rather, ifthe Uist empty, then the culprit ss whatever it Sitting at the top ofthe st anything. HF the (st 1s not empty = that iF arereka something farger tan the can to be moved on the target peg ~ then It tsastumed to be the obstractor (steps T3 and TA). “Table 1.2 shows the move selected by each of the partial strategies for each of the coven prem types along the tower-ending path from state 15 to stte t 1. GHILOREN'S PROBLEM SOLVING 197 Sp pt the cot an ee on [A STUDY OF CHILDREN'S PERFORMANCE se ty wen ono ist, wh eso ozo et nd Sy a eae wo osetia ae Sera ena th oa embed thea. The nnn eens Pee se ge pos dan the ion! of men BY periment Subjects 198 KLAHR 5:9, range 5:2 10 6:3). The sex ratio was approxnmately 50/50 in cach age range, ‘The children came from predowninantly ~ but wot excluswvely ~ upper-middle. class professional fanies. Phase 1: Purpose and Procedure ‘Those were thece goals for the ftst part of the study, which used the 4-yearold group: (1) Explore the basic ability of uninstructed 4.yearold children to solve problems of various lengths; (2} explore the extent co which they could describe PNnultple move sequence: and (2) explore the effectiveness of some rudimen: tary mstruction, meloding & graduated sequence of problems and a few supe hints about goals and subgoats “The child was faniliarved with the materls shown wv Fig. 7.2. Then the Jes and objectives were described 1 the context of a cover story. The cover story went someting like (hs ‘Once upon a tums there was 4 blue river {pom to space between rows of pups] On your side of the rver there were three browe teees, Can You Count your tres? On my side there were also three brown trees. On your Sie there lived three monkeys: 3 bys yellow daddy (pat yellow can om a mediom-ived blue mommy, anda Bitte red baby. The monkeys Tike to vamp fram tree to tree [according to the rues! they live on Your fide of the rivet. On my sie there are als three: a daddy, [ete Mine are Copyeat moneys. They want fo look just ike yours, ight across the river from your, Yours ave all sched up like so [state 11, anane are UKE 90 state 2 of 211. Mine are very unhappy: ean You tell me what (0 do so fine va Hook like your? [Theaetualseript nf cose, more elaborate ‘he problew sequence was generated by choosing wereasigly longer problems from alternating ses of the (wo tower-ending paths terminating mn state 1 (Fig. 753), for example, 21-1, 3-1, 1-1, 5- F,and so on, Ifthe cid suggested an iiegal move, the experimenter would point out the illegality. Hf the child had no suggestion for a move, the eaperuienter would make the correct fist move and then ask the child (0 continue. IF the child was sucessful at completing the problem in the rmimum number of moves, the experimenter wouldgive hint Either another problem of the same length from the other side of the state pace triangle or one that was cme move fonger. The idea behind all ofthis was to get an estimate of the upper performance level af the ebild without either having the child pive up because of too many falures or generating hicky sotutions. 'As the chi appeared to react his upper limit, the experimenter would begi to pe a systematic senes oF hints, altemptmg to move the child through the steps of the SOLVE strategy described earlier. For the fist Few problems, the children were asked to state not yut the next inoue but the next several moves. Children vaned widely an thew ability £0 do 1 cHonen'srnosLen soLvinG 199 TABLE? pare of Init! PF Stats or Prem of Oi ih nat Tork Prolers cia ogi er , CHILOREN'S PROBLEM SOLVING 203 hiven who could sole up 10 siemose, (OMe 08 problems, only one cette the Five moves atending reblm: 288 Ae nul solve only fout co tat-ending problems No tit could sole Aatending problem beyond mnt peronding Teel, and most of them dropped 4own fevels. Table 7.5 ey rept atthe 1 ents 0 Ne ve! righthand quadrant ore sears, the others ae a3 years of Be TNE results are consistent with ae par models m 2 weak singe, Te caf Ds demonstrated that for problems the aang 2, ae aval models aks nonce legal moves on Mat bev before they make any evors on tower PROB of the same length celine a petorance as path length aereasa WO regular, as shown soe at line Tale 7-5 sisting the conan Poesy that a subject wave an eve probtem wil fall am # ‘problem, OF the 30 win

You might also like