Вы находитесь на странице: 1из 17
coon scan 5, 11-82 Categorization and Representation of Physics Problems by Experts and Novices* MucnELeNe T. H. Ci Paul J. Ferovicn Rowexr Grass Univers of Peres Decree hi tom nao the pnt Co mete norton mae or soma ier preien ction «bons rene) rot {ne comprer ed by ener ood nna (2) ra bende ‘Socted with tg ond) ere nt pein a cnn ‘rien cegarztn and ropes, ead om sing top> ‘Set eal hat epr snd nv ap ppt ih [ete tet pti agro emptor ema on to Lnwiegn ocr wih he compen Far he pe iy caro pics pp wo eppromth cn oke 0 rb open ‘on, whercr oat oa penton ad appro ne oon ‘CATEGORIZATION AND REPRESENTATION OF PHYSICS PROBLEMS BY EXPERTS AND NOVICES ‘This paper presen stds designe to examine diferences inthe ways expert and novice problem solvers represent physics problems and to investigate imp ‘tions ofthese dfleencs for problem soltion. A problem representation i & Tic rah og, cnc ote Leming Ress ti Domest Cet supe a pet ene Ne, NOD CTS NR Se Oot Nl each ‘Sti atby eR tne of non, Poros se were ey ‘torte mec of be Aner Besa each Acros Ss rece, A 9 nt te meng a be Peto Soaey Pec Nove 17, Tees oe at {ee hep Aten, Ted Re, nd Capes Ra cmt econ, ‘tui aden. We pat spe potty oe ps pean sd ps ‘senso eed ete, nc) Red 8, Vane, Lari ce cal ‘ns fae bus niga commen uy Far Repti us at {pM Gag ewe Dp er Ce Pe Pt a ‘cognitive srsctarecomesponding toa problem, constructed by a wlver 00 the ‘ass of his dmain-eltd knowlege ad is organization. A representation an take avait of forms, Green (1977, for example, has proposed the represent tin of a problem a consructed semantic net conning virions compo- ‘es, Some ofthese comespond closely wid the problem as state, ich he {nial tate (the "ives", the desired goal, and te lgal problem-solving ‘operator (Newell & Simon, 1972) In aan, a represenaion can contin Chbelshment, inferences, and abstractions (Helle & Greene, 1979). Soe ‘belshmeat i on way of judging a slver's “understanding ofa problem (Green, 1977 ts posible tat with increasing experince ina domain, he ‘epreenation becomes ore etied. Te research descebed hee explores the ‘Sanger in problem repesctstion that emerge at a result of developing beet mauer expertise. Tris well Known by sow hat the quality of a problem representation iatcncs th ese with which a problem canbe solved (Hayes & Simon, 1976, ‘Newell Simo, 1972). In pies, Simon and Simon (197) have aribted he ‘xper's pga! ition” to the quality ofthe problem representation. The ‘uen conenss ir that the experts representation is superior because i com tains ret deal of golive knowiedge, De Klee (1977, fr example, bas Invedaeed both “"qonttatve” and "quale componcas inthe experts ‘eprecoation of « phyics problem where the qualitative component includes ‘oumathematal semantic exeripton of pial objects nd thei interactons. ‘Noval’ (1977) program ISSAC ako suggested some characterise of quai five presentation. In his program, pysieal objects from a problem salement lr epeseted not literally, bot ther, a brace bjetcteories—canonicl ‘bjt frames—each of which serves an equivalent physics role ©. pivot, lever or pint mast), The canonical abject frame isa knowicdgestrvtare th figment the information aboot an object stated in a problem with associated information from te Knowle base. In is ler work, Novak has roped the Inclsio of problem type in xtgorzation by pes (Novak & Araya, 1980). Categorization of «prcblem a8 ype would cue associated infomation in the ‘knowledge base, Silly, Reif (1979) as proposed a problem-solving mode in ‘which a ital step ia representation of “edscipon of ay problem inerms (of concepts provided by the Knowledge base" (p. 1). Tis knowledge base is ranged atund “problem schemata,” each of which contains information tecestary to solve 2 apeciic category of problems "The hypothesis puiding the pesca research stat the representation 1 ‘omircted i the context ofthe Knowledge avaliable for a particular type of problem. The knowledge wefl fr particular problem indexed when a given Dynes problem is categorized as a specie type. Thus, exper-aovie di- Ferwces maybe flat o poorly formed, qualitatively diferent, or monexistent Cegris in the novice representation I geocral, ths hypotesis is consistent ‘vith te perepmal chunking” hypothesis for experts (Chase & Simon, ‘ATEGOREANON A REREENTATON vm 1973 and its more general copie ramifications (Chase & Chi in pres), ‘which suggest hat emoch of expert powers inthe expert's ability to quickly ‘Stblsh conespondence between exeraly presctedcvets ad internal mod- efor thse evens "More pric, some evidence aleay exist inthe Iteratre Sugeest that solvers represent problems by caepey ad hat these categories may diet ‘problem solving. int, Hinsley, Hayes, and Simon (197) found tat college Staats can ctgorize algebra word problems into types, and tha his categor- {ation can ocr very quickly, somtimes even afer reading jst the fit phrase ofthe problem stemeat. For example, if subjects were to her the words ‘iver steamer then they might amis thal he problem Wat one shout cet ‘athe comparing the rates of going upsueam and downsteam. The ability 10 Enegoie problems quickly sagged to Hiney etl (1978) that “problem Shemata" eit and canbe viewed a interelated es of knowicdge that unify ‘Sapeafically dapat problcas by some underlying fetes. Secondly in chess ‘ecach it pers ta experts soperonity in memorizing chessboard positions les from the existence of & large sore of intact and well-organized cess oafiguations or pteras in memory (Chase & Simoa, 1973). Iti plausible that {choice among chess moves (enlogout to physics soliton methods) ress from «diet asocaion between move Sequeaces and a configura chunked) representation of the surface features of the board. Finally, ffom research in ‘medical dagnnis, there is evidence to suggest that expert dlgnontcan rep Sent particular caes by general categories, ad tht hese categories facia he fomation of hypotheses during diagoosi Pope, 1977; Wortman, 1972) "The accumulation of evidence forthe importance of categorization in ex. ‘per protiom solving leds ur to examiae the role of categorization in expert ‘Physics proba solving: particulary, investigate the relationships between ‘ich categorization and subsequent atemps at solatcn. The following seis of Stade temps to determine the catego hat expert and novices impose 08 piyscs problems (Sides One and Two); the knowledge which these categorical Feprescaatons activate ia the robe slver(Sady Thee); ad the cues ¢ features of probleme which subjects use to choose among aliemativ clegories (Study Fou, ‘Study One: Problem Sorting “The objective of theft stay was determine the knds of categories subject (ol ciflerat experience impose on problems. Using a soing procedure, we {ke cigh mfvanced PRD stents from te physi department experts) and ‘it undergaduts(9ovces) who ha just completed a semester of mechanics, to categorie 24 problems selected from Halliday ad Resnick’ (1974) Funda: Inewtols of Physics, beginsing with Chapter 5, Prcle Dynamics, and ending ‘Tih Chaper 12, Equlibiom of Bodies. Tce problems were selected from ry 4 raven ane ousen cach chapter, nd these were individually ypecon 3% S cade nations were ‘osertthe 24 problems into groops based on smarts of solute. The subjects wee not allowed to use pene and paper aod, ut, could no actly ale the problems inorder to sort them. As 2 tet of consistney, subject wer asked to Fear the problems aftr the Fst wal. Following this, dey were asked 10 ‘xplin the reasons for thls groupings. The ine taken 10 tort each ial was ‘aap measured. ‘Analy of Groot Quantitative Reslts [No gross quaniatve difereaces between the sors produced by the tw all oupe were observed. There were no differences i the number of ctegres produced by ech group (8.4 forthe expe an 8 for the ovies), ad the four largest categorie proce by each subject caprured the majority f the problems (0 perceat forthe experts and 4 percent forthe novices) Likewise, experts and ‘novices were equally able to achieve a stable sot within the to til, tat thei second sort matched ther fst sot very clasly. This suggests that thei sorting potern was nota hoc, bt rather, was based on some meeting pecan “Thee were, however, some difereaces in the amount of time it tok experts and novices wo sr the problems. In fact, experts took logger (8 nates ‘4S seconds per problem, on the average) to sr the problems in the fist tl (han novices (12 minutes or 30 second). Both groups were relatively fast at sorting the second wal (4-6 miaues forthe expert ad 5.5 minutes forthe soviet). The sped with which the probleme were sated onthe second tal (aout 12 secoads per problem) sugges that ejects probably dd ot vet g0 through the ene process of “understanding” each problem agai. Since the ‘problems were all aegviod afer theft rl the mbjets probably needed ‘aly to cat the ces that eliced eatgory members. In genera, thse uantiatve data sugest tat both experts and novices were abletcazgoie problems ino poops ina meaningfl way. Oter han the ifercoce inthe ime taken to sort on the first wal, there was ite difeence ‘between sil groupe. The eral question thea bocomes what are the bases on ‘which experts and noviescaepoie these problems? (Qualitative Anaiyis ofthe Categories Anaiyes of Four Pairs of Probleme. A cluser analysis (Dimer method) was performed on te problems grouped together bythe experts aad ‘tose bythe sovies. Such an analysis shows the degree to which subjects of ‘ach sil group ages that cern problems belong wo the same group. One way to interpret the cluster analysis is to examine oaly those problems tht were ‘ouped together withthe highest degre of agreement among subjects, (Our ina analysis centered on four pais of problems. Figures 1 and 2 contain the diagrams of pairs of problems that were grouped together by the ‘sovices and the expe, respectively. These dagrams canbe drawn to depict he physical situations described inthe problem statement, and we saneties given along witha problem statement (although no diagrams were given the ubjots ‘in ou studies). All eight movies grouped the top pur (Figure 1) together, a seven of the ight novices grouped the batiom pei. Both pais of problems (Figure 2) were grouped logeter by sx ofthe eight exper. "Examination ofthe novice pairs (Figure 1) reveals cerain sina in the surface structures ofthe problems. By “wuface structs," we meas: (4) dhe objects refered toi the problem (a sping, an inclined plane); (0) the tera ‘Physics terms mentioned inthe poble (fiction, center of mas) 0 (te ‘Physical configuration descibed in the probiem (.., rations among physical ‘objets suchas a block on an incline pene). Each pir of problems in Figure 1 ‘conains the same object components and conigurions—sicalar dks inthe ‘Upper par, blocks onan inclined plane in he lower pa, ‘The suggestion that movies eatgoriae by surface stuctue canbe con firmed by examining sbject’ verbal descriptions oftheir ctegoris (Samples ac given inthe figures.) Basically, according othe explanations, the op Pair ‘of problems involves “routional things" andthe bottom two problems “‘oks en inclined plans.” To reterate, the novices" wie of wurfice fates may involve either {keywords given inthe problem statement or abstracted visual configuration, at {in the presence of identical keywords (such scion is oe erterion by which novices group problems as sina. Yet, novices were also capable of going ‘beyond the word level to clauify by types of phyial objects: For example, “ery-goround" and “roaing disk” ae clsiid asthe se objet, a is the case forthe top pair of problems in Figure 1. For experts, srface features do not sem be the bases for categorization, ‘Theres weiter great snlary in he Keywords ured nthe problem semen, ‘or visual similarity apparent inthe digrams depicable from each pi of pro [ems shown in Figure 2. Nore the superficial appearance ofthe equations at ‘an be used on these problems the same. Only «physicist can detect he snl iy undeying the exper’ categoizaion. It appears tht the experts easly according tote major physi princple governing the solution ofeach problem, ‘The top pair of problems in Figure 2 can be solved by applying the Law of (Conservation of Energy; the boom pair is beter soived by applying Newion's ‘Second Law (F = MA). The vera jstification ofthe expert subjects confirm ‘tis analysis. If “deep soucture” is defined the warping physics lw ‘pliable toa problem; then, leary, hs deep structure isthe basis by which experts group the problems Analy of Categories, Further insight into the ways subjects categorize problems is given by the descriptions subject gave far the clegois they

Вам также может понравиться