You are on page 1of 32

Lawrence Erlbaum Associates (Taylor & Francis Group)

Discovery Learning and Transfer of Problem-Solving Skills


Author(s): Mark A. McDaniel and Mark S. Schlager
Source: Cognition and Instruction, Vol. 7, No. 2 (1990), pp. 129-159
Published by: Lawrence Erlbaum Associates (Taylor & Francis Group)
Stable URL: http://www.jstor.org/stable/3233474
Accessed: 19/09/2008 11:43

Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.

Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=lebtaylorfrancis.

Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.

JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.

Lawrence Erlbaum Associates (Taylor & Francis Group) is collaborating with JSTOR to digitize, preserve and
extend access to Cognition and Instruction.

http://www.jstor.org
COGNITION ANDINSTRUCTION, 1990,7(2),129-159
? 1990,Lawrence
Copyright Erlbaum Inc.
Associates,

DiscoveryLearningand Transfer
of Problem-SolvingSkills
Mark A. McDaniel
PurdueUniversity

Mark S. Schlager
SRI International

A frameworkfor understandingthe effectsof discoverylearningon the trans-


fer of problem-solvingskill is presented.A distinctionis drawnbetweenap-
plying a learnedstrategyon a transferproblemversushavingto generatea
novel strategyto solve a transferproblem.The main premiseof the frame-
workis that requiringdiscoveryof a strategywhilein trainingencouragesthe
activationor refinementof proceduresthat are useful for generatinga novel
strategy.In general,then, the primarybenefitof discoveryis that it shouldfa-
cilitate transferto tasks requiringa novel strategy.Learningby discovery,
however,may providelittle benefit for tasks that can be completedonly by
applyingthe learnedstrategy.Twoexperimentsprovidedsupportand further
refinementof this hypothesis.Experiment1 used a transferproblemthat
could be solvedwith the generalstrategylearnedin trainingbut requirednew
move sequencesto instantiatethe strategy.The resultsindicatedthat, when
transferrequirednew movesequencesto implementa generalstrategylearned
previously,discoverydid not enhancetransferof that strategy.In experiment
2, some transferproblemsrequiredusinga strategyotherthan that learnedin
training.As predicted,in this transfersituation,havingto discovera strategy
while in trainingproducedbettertransferthan beingprovidedwith a strategy
in training.Thus, discoveringa strategyprovidedbenefitswhen a new strat-
egy had to be generatedto solve a transferproblembut not whenthe learned
strategycould be applied,albeitwith newmoves,to the transferproblem.Ed-
ucationalimplicationsare discussed.

For the past three decades, psychologists and educators have engaged in
considerable debate concerning whether or not learning by discovery pro-
duces efficacious learning outcomes. Many educators have argued that dis-
covery learning produces better learning than expository learning (e.g.,

Requestsfor reprintsshouldbe sent to MarkA. McDaniel,Departmentof Psychological


Sciences,PurdueUniversity,WestLafayette,IN 47907.
130 McDANIELAND SCHLAGER

Bruner,1961). Othereducators,however,have expresseddoubts about the


benefits of discoverylearning(e.g., Friedlander,1965), and some have fa-
vored expositionover discovery(Ausubel, 1964; Sweller& Cooper, 1985).
An appeal to the literaturehas done little to resolvethis controversy,be-
cause there appearto be studiesthat supporteach view (McDaniel, 1980;
Scandura,1964).In this articlewe presenta theoreticalframeworkthat at-
tempts to better pinpoint the consequencesof acquiringproblem-solving
skill throughdiscovery,and we reporton two experimentsthat weredone to
examinethe utilityof this framework.
Ourgeneralapproachis basedon the information-processing analysesof
problemsolving that have proliferatedin the past decade. These analyses
have emphasizedthat proficientproblem-solvingperformancedependson
the developmentof appropriatecomponentskillsand knowledge(e.g., Reed
& Saavedra,1986).Webelievethat a majorshortcomingof extantcharacter-
izationsof discoverylearningis that distinctionsamongthe knowledgecom-
ponents underlyingskillful problem-solvingperformancehave not been
explicitlyconsidered(see Shulman& Keislar,1966).A furthercomplicating
factoris that the requisiteknowledgeand skills for problemsolvingappear
to differ acrossproblemtypes (Greeno,1978). For purposesof simplifica-
tion in this study,we focus on well-definedtransformationproblems:prob-
lems in which there is an initial situation, an explicit goal, and a set of
operations(i.e., moves)that transformone situationinto another.
A furthersimplifyingfeatureis that the study is limited to laboratory
transformationproblems,or "puzzleproblems,"not associatedwith rich
domain-specificknowledge.This disadvantage,however,is also the advan-
tage of these problems.The knowledgeunderlyingskilled performancein
these puzzle domainshas been well documented(e.g., Jeffries, 1979;Mc-
Daniel, 1980; Schlager,1984);there are reliableand objectivetechniques
for evaluatingthe kindsof solutionsattempted;and the amountof training
time neededto developexpertisein these problemdomainsvia discoveryis
not prohibitivefor researchpurposes.These propertiesalloweda more de-
tailedtheoreticalspecificationof the outcomeof discoverylearning(for the
problemdomain at hand), selectionof transfertasks that weretailoredto
test particulartheoreticalpredictions,and comprehensiveanalysesof per-
formance.Thus, we used laboratoryproblemshere in the same spirit as
other contemporaryresearchersin our field-as a way to developa viable
frameworkfor a complexinstructionalissue, a frameworkthat may prove
fruitfulin mountingan assaulton real-worldtasks (cf. Weisberg,1988).

IN SOLVING
COMPONENTSINVOLVED
KNOWLEDGE
PROBLEMS
TRANSFORMATION
We distinguishamongtwo knowledgecomponentsinvolvedin skilledsolv-
ing of transformationproblemsand a thirdcomponentinvolvedin generat-
DISCOVERYLEARNING 131

ing novel solutions. One knowledge component appears to include


strategiesthat specifya sequenceof subgoalsthat must be accomplishedto
solve the problem(Greeno,1978;Jeffries, 1979).A secondknowledgecom-
ponent is understandingthe particularmove sequencesneeded to accom-
plish the subgoals specified by the strategy.As a concreteillustrationof
these two components,considera problemused in the presentstudy. The
probleminvolves 5 hobbits and 5 orcs that must be transportedacross a
riverwith a seriesof boat trips, with the boat holdingno morethan 3 crea-
turesat a time. The majorrestrictionin the problemis the following.When
hobbits and orcs are together (on either side of the river bank or in the
boat), the orcs cannot outnumberthe hobbits.It is permissible,though, to
have only one type of creatureon eitherside of the riveror in the boat. A
strategythat experiencedsubjects appear to develop (Jeffries, 1979) in-
volves achievingthe subgoalsof movingall or nearlyall of the orcs across
the riverand then interchangingthe positionsof hobbitsand orcs (hobbits
moved across the river with orcs returningto the original side with the
boat). To achieve each subgoal, a sequenceof moves is required.So, in
additionto learninga strategy,subjectsalso must learnabout how individ-
ual movescan be combinedto satisfy the subgoals.Note that thereare sev-
eral differentmove sequencesthat can be used to accomplishthe subgoals.
Thus, although related, the strategy and the particularcombination of
moves that instantiatethe strategyare somewhatdifferentcomponentsof
expertise(i.e., one can learn the generalstrategywithout knowingthe ap-
propriatemove sequence,and, similarly,one can learn the move sequence
withoutlearningthe generalstrategy).
The thirdcriticalcomponentof problemsolvingidentifiedin our frame-
work deals with the generationof novel solutions. In some situations,the
problemsolver has either not yet acquiredknowledgeabout domain-spe-
cific strategiesand move sequences(the novice) or the problemsolver is
faced with a problemfor which learnedstrategiesand/or move sequences
arenot applicable.Consequently,in thesesituations,proceduresto generate
new strategiesand to choose movesmust be recruited.

THE EFFECTSOF LEARNINGBY DISCOVERY

In our view,the distinctiveoutcome of discoverylearningis the following.


During training,in the course of attemptingto discoverstrategiesand/or
the move combinationsthat implementthe strategies,the learneralso de-
velops proceduresthat help to generatenovel solution strategies(strategies
not acquiredin training)and/or novel move combinations.For example,
such proceduresmay includethe consciousprocessesinvolvedin activehy-
pothesistesting, that is, processesimplicatedin discoveryin some problem
domains (Lewis& Anderson, 1985). Further,we proposethat such proce-
132 McDANIELAND SCHLAGER

duresare specificto the particularkind of knowledgethat the learneris re-


quiredto discoverin the trainingcontext;this proposalis consistentwith
recentfindings and theoriesof skill acquisition(e.g., see Kolers& Roedi-
ger, 1984).Specifically,procedureslearnedfor discoveringthe movecombi-
nationsnecessaryto implementa generalstrategy,providedduringtraining,
will be effectivefor discoveryof other move combinationsfor implement-
ing that strategy.Theseprocedureswill not, however,be especiallyeffective
for discoveryof a novel strategy,whichwouldbe neededto solve a transfer
problem. Similarly,procedureslearned for discoveringa strategyduring
trainingwill be effectivefor discoveringa novel strategyto solve a transfer
problem,in the domain. By this view,any advantageof discoverylearning
wouldbe limitedto those situationsin whichthe transfertask requiressub-
jects to discoverinformationof the type that had to be discoveredfor suc-
cessful performanceon the trainingtask.
Thesenotionscan be distinguishedfroma "commonelements"approach
to transfer(Thorndike& Woodworth,1901). The common elements ap-
proachsuggeststhat stimulus-responseassociationsacquiredto performan
initialor trainingtask producetransferwhenthey can be applieddirectlyto
a subsequenttask. This positionhas beenembracedand furtherspecifiedin
contemporaryproblem-solvingliterature(cf. Brown& Kane, 1988;Polson,
1988),and it has gainedpredictiveand explanatorypowerin Polson's(1988)
framework,in which common elementsare assumedto be rules used to
achievethe same goals across differentproblemsin a domain (e.g., com-
putertext-processingtasks). As appliedto the computertext-processingdo-
main, the idea is that rules used to achieve goals that are common to
differenttext-editingtasks are incorporateddirectlyinto solutionsto trans-
fer tasks (i.e., the rule is incorporatedinto the solutionof the transfertask
exactlyas it was used in the originaltask). Extendingthis idea to the current
problemdomain,the commonrulesmightbe thoughtof as movesor move
sequencesthat arecommonto the trainingandtransferproblems.According
to this view, discoverylearningmight enhancetransferbecause discovery
would lead to explorationof particularmove sequenceson the training
problem(s)that could be directlyintegratedinto the solutionsrequiredon
the transferproblem.By contrast,in our view,enhancedtransferfor discov-
ery learners,whenobtained,will be primarilydue to moreefficaciousselec-
tion of new move sequencesand strategiesrequiredto solve the transfer
problem.Again, the idea here is that discoveryfostersproceduresthat are
useful in generatingnew move sequencesand/or strategies.

EMPIRICALPREDICTIONS

Our frameworkleads to the followingworkinghypothesisregardingthe ef-


fectivenessof variousinstructionalsituationsthat differ with regardto the
DISCOVERY LEARNING 133

degreeof discoveryrequired.In a given problem-solvingdomain, Instruc-


tion A will benefit transferrelativeto InstructionB only when Instruction
A requireslearnersto discoverinformationduringtrainingthat is from the
same knowledge-component level as the informationthat has to be discov-
eredto completethe transfertask and when InstructionB does not require
discoveryof such informationin training.This hypothesisleads to two spe-
cific predictionsthat are testedin this article.In statingthe specificpredic-
tions, it will be helpful to use labels to distinguishbetweentwo possible
instructionalsituationssuggestedby our analysis.In one instructionalsitu-
ation, learnersare providedwith a directstatementof the solution strategy
but arerequiredto discoveran appropriatemovesequenceto instantiatethe
strategy;we will label this type of instructionmove-discovery(MD). In an-
other instructionalsituation, learnersare requiredto discoverthe strategy
and an appropriatemove sequence;we will label this type of instruction
strategy+ move-discovery(SMD).
SMD instructionwill enhancetransferrelativeto MD when the transfer
problemsmust be solved with a differentstrategyfrom that used to solve
the trainingproblem(Prediction1). SMD instructionwill producetransfer
performancethat is indistinguishablefrom that of MD instructionwhen
the transferproblemscan be solved with the same strategylearnedduring
instructionbut requirea differentmove sequence(Prediction2).
The literatureprovidessome preliminarysupportfor this predictedpat-
tern. Forexample,givenmatchsticktransferproblems'that could be solved
with the strategylearned during training, but requiringdifferent moves,
SMD subjectstransferredno betterthan MD subjects(Corman,1957). (In
training,MD subjectswereprovidedwith a strategystatement-"Matches
shouldbe movedso that no squarehas a commonside with anothersquare.
Everymatchshould be the side of only one square"-butnot the particular
moves.) Waters(1928), using bead problems,and McDaniel(1980), using
river-crossing problems,also failedto find significantdifferencesin instruc-
tional groups similarto the ones under considerationhere on transferto
problemsrequiringapplicationof the strategylearnedduringtraining.
Results consistentwith both predictionshave been reportedin a non-
transformationalproblem domain (e.g, cryptogramproblems). Subjects
who had to discovera cryptogramrule did not transfersignificantlybetter
than subjectsprovidedwith the rule in acquisitionto problemsfor which
the rule applied. But subjectswho had to discoverthe rule in acquisition
transferredsignificantlybetter than those subjectswho were providedthe
rule to problemsfor whicha new rule was required(Guthrie,1967).
The presentstudy reportson two experimentsthat implementedthe dis-

'Matchstick problems consist of a number of two-dimensional shapes (e.g., squares) that


can be built with matchsticks. One particular kind of problem requires that the number of
shapes be increased by one by moving no more than X matches (e.g., X might be 3).
134 McDANIELAND SCHLAGER

covery-learningconditionsidentifiedin the foregoinganalysis.Experiment


1 testedtransferwith a problemthat could be solvedby applyingthe strat-
egy learnedduringtraining.Buildingon the first experiment,Experiment2
testedtransferon a problemthat requiredgenerationof a strategydifferent
from that learnedin training.Additionally,the second experimentyielded
empiricalanalysesthat made it possible for us to distinguishbetweenthe
frameworkdevelopedhere and a common elementsapproachto discovery
learningand transfer.

EXPERIMENT1

In this experiment,we testedpredictionsdesignedto accountfor the situa-


tion whenthe transferproblemcould be solvedwith the samegeneralstrat-
egy used to solve the trainingproblem.The strategywas acquiredthrough
one of threeinstructionalconditions.In the SMD condition, subjectswere
requiredto discoverthe generalstrategyand the moveson their own while
solving a hobbits-orcsriver-crossingproblem(describedin the preceding
section). In the MD condition, subjects were explicitlytold the general
strategythat could be usedto solveany variationof the river-crossing
prob-
lem. Then these subjects were left to discover an appropriatemove se-
quenceto instantiatethe strategywhile solving a hobbits-orcsproblem.In
the no-discovery(ND) condition, subjectswereexplicitlytold the strategy
and a descriptionof a move sequencefor instantiatingthe strategy.These
subjectsthen practicedthe strategyand associatedmoveson a hobbits-orcs
problem.ITo weeks after training,subjectssolveda transferproblemthat
could be solved with the same strategyused on the trainingproblem,but
the transferproblemhad somewhatdifferentconstraintsand requireddif-
ferentmovesfrom those learnedduringtraining.
Our frameworksuggests that, for the present transfercontext, SMD
trainingwill confer no benefit in transferrelativeto MD training(Predic-
tion 2). Decrementsin transfershould be limitedto ND training,because
this condition does not have to developproceduresto generatenew move
sequencesto instantiatethe learnedstrategy.

Method

Design and procedures. A between-subjects designwith threelevelsof


instruction-SMD, MD, and ND-was used. All subjectsweretrainedon a
hobbits-orcsriver-crossingproblemin which 5 hobbits and 5 ores had to
cross the riverin a boat holdingno morethan 3 creatures-labeleda (5, 3)
problem.In the SMD condition, subjectsweresimplyinformedabout the
constraintsof the problemand were then requiredto solve the problem.
DISCOVERY LEARNING 135

These subjects receivedno instructionon either the generalstrategyor a


particularmove sequencefor solvingthe problem.
In the MD instructionalcondition, prior to solving the (5, 3) training
problem,subjectswerepresentedwith a strategyfor solving any variation
(i.e., differencesin numberof creaturesor boat size) of the hobbits-orcs
problem.The strategystatementreadas follows: "First,move all or nearly
all of the orcs acrossthe river.Then, interchangethe positions of the hob-
bits and orcs. Finally,again transportthe orcs acrossthe riverto complete
the problem."This strategywill be labeled orcs-switch-orcs(OSO), after
Jeffries(1979).
The ND groupwas givenexplicitinstructionon the generalOSO strategy
and on a particularmove sequencethat could be used to solve the (5, 3)
hobbits-orcsproblem.This instructionstated:

In doing the task, the followingstrategyis used for solvingthe prob-


lem: First, move all or nearlyall of the Orcsacrossthe river:Trans-
port three Orcs to the far bank and returnthe boat with an Orc;
transporttwo Orcsto the far bank and againreturnthe boat with an
Orc.
Then interchangethe position of the Hobbits and Orcs:Transport
three Hobbits to the far bank and returnwith a Hobbit-Orc pair;
transportanotherthreeHobbitsto the far bank.
Finally,again transportthe Orcs across the riverto completethe
problem:Returnthe boat with an Orc, transport3 Orcs across the
bank, returnwith an Orc,and transportthe remaining2 Orcsto com-
plete the problem.

For purposesof clarification,we now providea brief descriptionof how


the OSO strategyrelatesto particularmove sequencesthat solve the prob-
lem. Figure1 depictsa seriesof legal movesfor the (5, 3) problem.The rec-
tangle at the top of the figure representsthe start state (i.e., 5 hobbits, 5
orcs, and a boat, all of whichare positionedon the left bank of the river);
this is State 2 in Figure1. The rectangleat the bottom of the figurerepre-
sents the goal state;this is State27 in Figure1. Focusingat the top half of
Figure 1, it can be seen that, to solve the problem, one must proceed
throughState 13. Thereare variousmove sequencesthat bringone to State
13, but all of these sequencesrequirethat all or nearlyall of the orcs cross
the river(i.e., the first goal encompassedin the strategystatement).After
arrivingat State 13, the next phase involvesmoves that bring the hobbits
acrossthe riverand returnorcs to the originalbank (leadingto eitherState
19 or 21). This move sequenceis reflectedin the goal describedby the sec-
ond strategystatement.Finally,the problemis solved by transportingorcs
acrossthe river(as per final strategystatement),again, a subgoalthat can
be accomplishedby severaldifferentmove sequences.Note that the mini-
START

GOAL

FIGURE 1 Problem space of legal moves for the hobbits-orcs (5,3) problem. The let-
ter H stands for hobbit, O stands for orc, and the symbol "< > " is used to represent
the boat; the spatial location of the symbols represents the bank of the river (eft or
right) on which the objects are located. The symbols placed between the states indicate
the move that transforms one state into another state.

136
DISCOVERY LEARNING 137

mum numberof moves in which the problemcan be solved is 11 and that


thereare variousmove sequencesthat can be used to accomplishthis mini-
mum-pathsolution, all of which can be consideredinstantiationsof the
OSO strategy.
After receivingthe assignedinstructionand solving the (5, 3) hobbits-
orcs problem to a criterionof three successiveminimum-pathsolutions
(i.e., 11-movesolutions), subjectsweredismissedand told to returnto the
laboratoryin 2 weeks.The trainingcriterionwas establishedto ensurethat
all subjectswouldacquirethe OSO strategy.
Upon returningto the laboratory,subjectswereaskedto solve the trans-
fer problem. The transferproblem was a "jealous husbands"problem,
whichcan be statedas follows:

Eight jealous husbandsand their wives having to cross a river at a


ferry,find a boat but the boat is so smallthat it can containno more
than five persons. Find the simplest scheduleof crossingsthat will
permitall 16 people to crossthe riverso that none of the womenshall
be left in the company with any of the men, unless her husbandis
present.It is assumedthat all passengerson the boat unboardbefore
the next trip, and at least one person has to be in the boat for each
crossing.(adaptedfrom Reed,Ernst,& Banerji,1974, p. 438)
This problemis similarto hobbits-orcs,in that the solution paths of the
two problemsare identical,providedthat husbandsare substitutedfor hob-
bits, wives are substitutedfor orcs, and the husbandsand wives are paired
as couples.After minimaltraining,however,subjectsdid not appearto per-
ceivethis similarity(Reedet al., 1974).Thus,the jealous husbandsproblem
appearedto presenta challengingtest of subjects'ability to transferthe
strategylearned during training. Specifically,this problemcan be solved
with no less than nine moves,but only by applyingthe OSO strategy;by ab-
breviatingthe movesequencethat instantiatesOSO,relativeto the 11-move
sequencerequiredfor the (5, 3) hobbits-orcsproblem;and by substituting
husbandsfor hobbitsand wivesfor orcs.
Subjectswererequiredto solve the jealous husbandsproblemrepeatedly
until they wereable to do so once in minimumpath. All subjectswho failed
to solve in minimumpath by Trial2 weregivenan additionalhint preceding
each of theirsubsequenttrials:

The solution of the JealousHusbandsproblemis the same as the so-


lution of the Hobbits-Orcsproblemif one substitutesa Husbandfor
a Hobbit and a Wife for an Orc.Wheneveryou moveda Hobbit pre-
viously, you should now move a Husbandand wheneveryou moved
an Orcpreviously,you should now move a Wife.
138 McDANIELAND SCHLAGER

The trainingand transferproblemswerepresentedon a cathode-raytube


(CRT)computermonitor.Subjectsenteredtheirmovesby depressingspeci-
fied keyson a keyboard;they proceededat their own pace.

Subjects. Originally,a total of 90 studentsfrom an introductorypsy-


chology class at the Universityof Notre Dame participatedto receiveextra
creditpoints. Randomassignmentof subjectsto conditionsresultedin 31
subjectsin the SMD condition, 30 subjectsin the MD condition, and 29
subjectsin the ND condition.However,one subjectin the SMD condition,
3 in the MD condition, and 5 in the ND conditionwerereplacedbecause
they either failed to returnfor the transfersession or their data were un-
availabledue to equipmentproblems.

Results
Training. For all analyses,the rejectionlevel was set at .05. Perform-
ance on the trainingproblemcan be simply summarized.As increasingly
more information about the solution strategywas providedto subjects
(from no informationin the SMD conditionto detailedinformationin the
ND condition),performanceimproved(see Table1 for means).This obser-
vation was supportedwith analysesof variance(ANOVAs)of trialsto last
error(i.e., numberof trials before three consecutiveminimum-pathsolu-
tions), legal moves, and illegal moves (for legal and illegal moves, mixed
ANOVAsthat includedthe first five trialsas a factorwereconducted).Sig-
nificanteffects of instructionwerefound for all threedependentmeasures:
For trials to last error, F(2, 87) = 4.20, MSE = 10.56; for legal moves, F(2,
87) = 12.94, MSE = 178.99; for illegal moves, F(2, 87) = 6.17, MSE = 9.14.
Also, performanceimprovedsignificantlyacross trials: For legal moves,
F(4, 348) = 25.59, MSE= 110.65; for illegal moves, F(4, 438) =43.42,
MSE= 6.47. Finally, significant Groupx Trials interactions-for legal
moves, F(8, 348) = 3.65, MSE = 110.65; for illegal moves, F(8, 348) = 3.19,
MSE = 6.47-indicated that differencesbetween the instructionalgroups
werereducedor eliminatedas subjectscompletedmoretrials.
Following Jeffries's (1978) technique,we identified subjects' solution
strategiesaccordingto the movesequencesthat the subjectsused in the ini-
tial part of the problem.We classifiedthe move sequencesinto one of two
basic strategies:a balancestrategyand an OSO strategy.The balancestrat-
egy was one in whichsubjectstriedto keepthe numberof hobbitsand orcs
as even as possibleon both sides of the river.This strategyresultedin sub-
jects movingfrom States5-7-9-12 (see Figure1). Also classifiedas balance
strategieswerethose instancesin which subjects'initial movesto the right
bankinvolvedmovingorcs into State3, afterwhichsubjectsmovedhobbits
to balance the orcs and continued trying to maintainthis balance with
movesequencesfrom States3-7-9 or 12, or 3-6-9 or 12. None of thesebal-
DISCOVERYLEARNING 139

TABLE1
Performanceon the FirstFiveTrainingTrialsas a Function
of InstructionalConditionin Experiment1

InstructionalCondition

Training7rial ND MD SMD

Trial1
Legalmoves 17.5 26.1 35.7
Illegalmoves 2.3 4.3 6.3
PercentageusingOSO 89.7 53.3 29.0
Thial2
Legalmoves 14.6 17.7 26.3
Illegalmoves 0.7 0.9 2.1
PercentageusingOSO 82.8 63.3 41.9
Trial3
Legalmoves 13.2 13.9 18.9
Illegalmoves 0.3 0.3 0.5
PercentageusingOSO 82.8 70.0 48.4
Trial4
Legalmoves 12.5 13.7 13.6
Illegalmoves 0.1 0.2 0.3
PercentageusingOSO 93.1 73.3 48.4
Trial5
Legalmoves 12.5 12.3 14.4
Illegalmoves 0.0 0.1 0.4
PercentageusingOSO 93.1 83.3 67.7
Numberof trialsbefore 2.4 4.0 4.8
minimum-pathsolution
Note. Subjectswho met criterionbeforeTrial5 wereassignedminimum-path
performance
for subsequenttrialsratherthanbeingdroppedfromthe analyses.

ance sequences leads directly to a solution of the problem; they lead to a


dead-end state (e.g., State 12).
Any sequence of moves that involved moving only orcs until State 13-
as described in the Method section, this marks completion of the first goal
of the OSO strategy-was considered an OSO strategy. Although this strat-
egy can involve backtracking, it does not have to. It can be used to solve the
problem in the fewest number of moves possible.
The strategy classification analysis confirmed that the ND group imme-
diately used the OSO strategy, the MD group was slower to adopt OSO, and
the SMD group was the slowest to use OSO (see Table 1). The differences in
strategy use were significant on Trials 1 through 5, X2(2,N = 90) ranged
from 22.63 on Trial 1 to 6.44 on Trial 5. After Trial 5, the percentage of
subjects using OSO was high and not significantly different across groups.
These findings converge with research on real-world problems, for in-
stance, Sweller and Cooper's (1985) work examining acquisition rates of al-
gebra problem-solving skill as a function of type of instruction. They found
140 McDANIELAND SCHLAGER

that worked examplesthat presumablydirectedattention to a "schema"


that was useful for solving a particulartype of problem(e.g., a particular
algebraproblem)producedmore efficient acquisition(less time and fewer
errors) than an instructionalcondition that requiredsubjects to work
throughpracticeproblemson theirown.

Transfer The groupsdid not differ significantly(F< 1) in the number


of trials taken before achievinga minimum-pathsolution to the jealous
husbandsproblem(3.5, 3.4, and 3.9 trials to last errorfor the ND, MD,
and SMD groups,respectively).Giventhat naive subjectstook an average
of 9.0 trialsto last erroron the homomorphic(8, 5) hobbits-orcsproblem
(McDaniel,Polson, & Beecher,unpublishedstudy,1979;citedin McDaniel,
1980), all groups in this study appearedto demonstratepositive transfer.
Two-factormixedANOVAsof legal and illegalmovesfor the first threetri-
als also revealedno significanteffects as a function of instructionalgroup
(Fs < 1; Table2 displaysmeans for the first two trials only, because per-
formancewas very similaracross instructionalconditions for all transfer
trials).An objectiveanalysisof the solutionattempts-using the procedure
and the three strategytemplatesdescribedin AppendixA-revealed that
the kinds of solutionsattemptedwerealso similarfor the differentinstruc-
tional conditions(see Table2).

Discussion
The equivalenttransferfor SMD instructionand MD instructionwas con-
sistent with the second predictionderivedfrom the frameworkpresented
earlierin this article. This framework,however,would also anticipatethat
subjectswho had to discovera move sequenceto instantiatethe requisite
strategy(i.e., SMD and MD conditions)would be better able to generate
new moves to apply the strategyon the transferproblem,relativeto sub-
jects who weregiven a move sequenceduringtraining(i.e., the ND condi-
tion). This hypothesiswas not supported.
The most strikingfinding was the good transferperformanceof all of
the groups,especiallyin light of the relativelylong retentionintervaland a
transferproblemwith surfacefeaturessomewhatdissimilarto the training
problems.Transferbetweenproblemisomorphsis characteristicallydiffi-
cult to obtain (e.g., Catrambone& Holyoak, 1989;Reedet al., 1974).Ac-
cordingly,a central challenge to problem-solvingtheorists has been to
explain the obstaclesto positive transfer(e.g., Bassok & Holyoak, 1989;
Novick, 1988;Reedet al., 1974;Stein, Way,Benningfield,& Hedgecough,
1986). One salient differencebetweenmost of the extantresearchand the
presentwork is the amount of exposuregiven to the trainingproblem.In
most other studies, subjectswereinstructedto solve the trainingproblem
DISCOVERYLEARNING 141

TABLE2
Trialsin Experiment
on FirstTwoTransfer
Performance 1

InstructionalCondition

Transfer rial ND MD SMD

Trial1
Legalmoves 12.7 12.9 13.2
Illegalmoves 1.0 1.0 1.1
Strategyclassification(in percent)
OSO 17 13 15
Paired 41 40 39
Mixed 24 30 26
Trial2
Legalmoves 12.6 11.1 11.4
Illegalmoves 1.0 0.5 0.4
Strategyclassification(in percent)
OSO 21 50 32
Paired 28 30 29
Mixed 28 20 26

Note. Combinedstrategyclassificationpercentagestotalingless than 100%indicatethat


some subjects'solutionscould not be classifiedwithinthe specifiedstrategycategories.See
AppendixA for descriptionsof the strategycategories.

only once before they attempted to solve a transfer problem. In the present
experiment, subjects continued their attempts to solve the training problem
until they performed the task perfectly. Perhaps one important ingredient
for effective transfer of a strategy or schema (Catrambone & Holyoak,
1989; Sweller & Cooper, 1985) is complete learning of the solution strategy
during training. Apparently, the degree to which training involves discovery
of the strategy is irrelevantas long as the strategy is well learned and as long
as the learned strategy is applicable for the transfer problem.

EXPERIMENT2

Our framework suggests that there is a context in which learning a strategy


well is not sufficient to ensure good transfer. Specifically, this context is one
in which the transfer problem cannot be successfully solved just by using
the strategy that solved the training problem. In this context, we suggest
that subjects must activate procedures for developing a new strategy. Ac-
cording to our framework, these procedures are sharpened, or acquired,
through SMD instruction but not through instruction that explicates the
strategy for the learner (i.e., the MD condition). Unfortunately, this predic-
tion is not testable in the river-crossing domain, because the OSO strategy
142 McDANIEL AND SCHLAGER

can solve any river-crossingproblem.2We thereforeselectedanotherprob-


lem domain-water-jar problems(Luchins,1942). Water-jarproblemsare
similarto river-crossing problemsin that they aretransformationproblems,
and the transformationsare effectedthroughmoves.The advantageof the
water-jardomainis that problemscan be selectedto exemplifya varietyof
uniquesolutionstrategies.
In a typicalwater-jarproblem,the subjectis askedto imaginea number
of emptyjars (usuallythree)on a counter,neara sink. Eachjar can hold a
specified volume of water but has no graduatedmarkingswith which to
measurefractionalamounts.The subjectis giventhe capacitiesof the three
jars and an amount of waterto be obtained.The object is to figureout a
sequence of filling and emptying operationsthat will result in the goal
amountresidingin one of the jars. For example,the subjectis told that Jar
A can hold 7 units, Jar B can hold 24 units, and Jar C can hold 98 units.
The goal is to obtain 60 units of waterin Jar C. To solve this problem,one
must first fill Jar C fromthe sink, then, using JarB, take out 24 units from
Jar C, leaving74 units.Next, JarA is used to take 7 units from Jar C, leav-
ing 67 unitsin Jar C. Finally,JarA is againusedto take 7 units from Jar C
(the originalcontentsof Jar A werefirst pouredinto the sink), resultingin
the goal of 60 units. Thus, the solution for this problemis given by the
equation 1C - 1B - 2A. This particularmove sequenceis an instantiation
of the more generalstrategy:Subtractone jar from anotherjar and, then,
subtract the third jar twice (1 - 1 - 2).
For presentpurposes,once the learnerhas mastereda strategysuch as
1- 1-2 (by solving severalproblemswith solutions that are particular
instantiations of the strategy, e.g., 1C - 1B - 2A, 1A - 1C - 2B,
1B - 1A - 2C), two kindsof transferproblemsare of interest:(a) problems
that are solvedusingthe samegeneralstrategylearnedduringtraining(i.e.,
1 - 1 - 2) and (b) problemsthat requirea differentgeneralstrategy,for ex-
ample,a strategythat tells one to subtractthe smallestjar from the largest
jar six times and, then, add the mediumjar (i.e., 1 - 6 + 2).
As in Experiment1, subjectsweretrainedto solve the initialproblemsin
the domainbeforebeing tested on new problems.Twoinstructionalcondi-
tions-SMD and MD-were implemented.This experimentextendedthe
previousexperimentby (a) requiringsubjectsto learn four differentstrate-
gies (along with the movesthat instantiatedthe strategies),(b) using trans-
fer problemswith a solution that involveda differentstrategyfrom those
learnedin acquisition,and (c) manipulatingthe acquisition-testdelay.

2Traininga balance (i.e., a paired strategy) strategy and transferring to a problem solved
only with OSO seems, at first glance, reasonable. Novices, however, are overwhelmingly biased
toward using a balance strategy (see McDaniel, 1980). Thus, one would not expect that explic-
itly providing subjects with the balance strategy would produce any differences in acquisition
relative to the SMD group, which was not explicitly provided with the balance strategy.
DISCOVERYLEARNING 143

As in Experiment1, we anticipatedthat the two discovery-instructional


conditionswouldproduceequallyeffectivetransferto a problemthat could
be solved using a strategylearnedin training,regardlessof acquisition-test
delay. For transfer problems solved with a strategy different from that
neededto solve the acquisitionproblems,however,we anticipatedthat the
SMD groupwould performbetterthan the MD group,whichwas provided
with the strategyduringtraining.In addition, the SMD advantageshould
endureoveran extendedtest delay.

Method
Design. The design was 2 x 2 x 2 between-subjectsfactorialincluding
the variablestype of instruction(MD, SMD), type of transferproblem
(near,remote),and test interval(immediate,48-hrdelay).All subjectswere
trainedon 16 three-jarwater-jarproblemspartitionedinto four sets; each
set exemplifieda differentstrategy.The first set containedfour problems
that were governedby a strategyof filling the largestjar, subtractingthe
volume of anotherjar twice, and adding the volume of the remainingjar
(1 - 2 + 1). The second set of four problemswas governedby a strategyof
filling the largestjar, subtractinganotherjar three times, and subtracting
the remaining jar twice (1 - 3 - 2 or 1 - 2 - 3). The strategy for the third
set involved filling the largestjar, subtractinganotherjar once, and sub-
tracting the remaining jar twice (1 - 1 - 2 or 1 - 2 - 1). Finally, the fourth
set was governedby a strategyof fillingthe largestjar, subtractinganother
jar twice,and addingthe remainingjar four times (1 - 2 + 4; see Appendix
B for the completeproblemset).
The MD subjectswereinformedthat they would be shown a strategyat
the beginningof each set of trainingproblemsthat would help them solve
all problemsin that set. The strategiesfor solving these problemstook the
form of mathematicalexpressionsin whichthe jar labelswereleft out. For
example,the strategy1 - 2 + 1 could be appliedto a problemin which Jar
C is the largestand Jar A is the smallest, Jar B the largestand Jar C the
smallest, and so on. Thus, these subjects had to discoverthe particular
moves to implementthe strategy.Subjectsweretold to take their time be-
fore startingthe first problemin orderto memorizethe strategy,becauseit
wouldbe presentedonly once. An exampleof the mathematicalformof the
strategies(i.e., as displayedearlier)was also provided.
The SMD subjectswerenot given the strategiesbeforeeach set but, in-
stead, wereinformedthat each problemin a set must be solved using the
samegeneralstrategy.Theyweretold the strategywouldbe a moreabstract
form of the specific equationsrepresentingthe solution of each problem,
and they werealso told to try to discoverthe strategyfor each set, because
it would help them solve the problemsmore easily.
Two types of transferproblemswere employed. One type represented
144 McDANIEL AND SCHLAGER

problemsthat could be solved with a strategylearnedin training.Thus,


these problemsare analogousto the transferproblemsemployedin Experi-
ment 1. The other type representedproblemsthat were not solvablewith
the trainedstrategies.Forease of exposition,we will label the formerprob-
lem-type as "near" (transfer)and the latter problem-typeas "remote"
(transfer).One set of two near-transferproblemsand one set of two re-
mote-transferproblemswereused.
One near-transferproblemwas a problemnot seen in training,but its so-
lution was governedby the 1- 1 -2 strategyfrom the third trainingset.
This particularstrategywas randomlychosenfromthe four taughtin train-
ing. This problemis labeledthree-jarneartransfer.Subjectswereinformed,
before the problemwas presented,that it could be solved by using one of
the strategieslearnedin the training.The othernear-transferproblemwas a
problemthat could be solvedusingonly two jars; thus, it also was not seen
in training.It was solvedby fillingthe largestjar and, then, subtractingthe
smallestjar (1 - 1). Although it does not representan exact instanceof a
strategylearnedin acquisition,instantiationof the just-mentionedstrategy
directlysolvesthe problem(i.e., fill the largestjar and subtract).This prob-
lem is labeled two-jar near transfer.Subjects were informed before the
problemthat it could be solvedby using only two jars.
The remote-transfer set also consistedof 1 three-jarand 1 two-jarprob-
lem. Neitherof these problemswas solvablewith the strategieslearnedin
acquisition. The three-jarproblem involved subtractingthe smallest jar
from the largestsix times and, then, addingthe mediumjar. In acquisition,
the most closely relatedstrategiesinvolvedsubtractingthe mediumjar and
addingthe smalljar. The two-jarremoteprobleminvolvedaddingthe me-
dium jar to the small jar (the contentsare collectedin the largejar). This
two-jarproblemcould not be solved by instantiatinga strategylearnedin
acquisitionbecauseall of the acquisitionstrategiesinvolvedfillingthe large
jar as the first operation.
Half of the subjectsreceivedboth near-transferproblems,and the other
half receivedboth remote-transfer problems(thethree-jarproblemswereal-
ways presentedfirst). The test intervalvariableinvolvedadministrationof
the transferproblemseitherimmediatelyor 48 hr aftertraining.Finally,two
controlgroupsthat solvedonly the transferproblems(one groupsolvedthe
near-transfer problems,andthe othergroupsolvedthe remote-transfer prob-
lems)wereincludedto gaugethe difficultyof eachproblemandto determine
if the trainingproceduresproducedpositivetransferor negativetransfer.

Subjects. One hundredforty students at the Universityof Colorado


participatingin partialfulfillmentof an introductorypsychologycoursere-
quirementservedas subjects.Fifteen subjectswere randomlyassignedto
each of the eight experimentalgroups, and 10 subjects were assignedto
each of the two controlgroups.Fourteenadditionalsubjectswereexcluded
DISCOVERYLEARNING 145

for failure to complete the training phase, 4 were lost in the control condi-
tion for failure to solve both problems, 3 were terminated by computer fail-
ure, and 6 were excluded for other reasons (1 due to illness, 3 chose not to
continue, and 2 due to extreme fatigue).

Procedure. Subjects were seated in front of a CRT terminal and were


asked to solve some problems on the screen, using the keypad. After a gen-
eral orientation to the equipment, the experimenter revealedthe first instruc-
tion screen and left the room. From this point on, the pace of the experiment
was controlled by the subject. Subjects were allowed to review the instruc-
tions as many times as needed and to call the experimenter at any time.
The instructions included general water-jar problem constraints, a
worked example, and detailed instructions on use of the terminal. Subjects
also had access to a sample problem screen and three examples of how
moves could be made by using the keypad. This information, in the form of
a booklet, was coordinated with instructions on the screen. The final in-
struction screen contained either the MD (i.e., instruction on the provided
strategy) or SMD instructions. All subjects were then given one last chance
to review the instructions. Once the task had begun, subjects were not al-
lowed to review the instructions or the strategy. Subjects were not permitted
to move on to the next problem until the current one was solved. To ensure
that subjects learned the strategies and did not solve the problems by
chance, they were required to enter the correct solution equation into the
computer before being shown the next problem. If, after two attempts, the
subject had not correctly entered the equation, the problem was automati-
cally started over.
Each move, as well as its result, was recorded on the screen to reduce
memory load. Subjects were able to clear the screen of moves and empty
all jars in order to start over if so desired.3 If an illegal move was made, a
message flashed on the screen saying, "Illegal move-Please try again."
This remained until a legal move was entered. Subjects were allowed an un-
limited number of moves. All subjects were given 1 hr and 50 min to com-
plete the training phase. If at this time a subject had not completed the 16
training problems, that subject was given credit and terminated from the
experiment.
After successful completion of the training phase, the no-delay groups
were presented with one of the two transfer-problem sets (i.e., near-transfer
or remote-transfer). The delay groups were told that the current part of the
experiment was over and that they were to return in 48 hr to solve a few

3Unlikeriver-crossingproblems,water-jarproblemsareopen ended.Thatis, thereis no set


patternof possiblemovestateswith well-defineddead ends and uniquesolutionpaths.Blind
alleyscan go on indefinitely,and retreatto the startstatecan becomea formidabletask in it-
self. Forthis reason,we allowedsubjectsto pressa keyand beginthe problemagainwhenthey
felt that a blindalleyhad beenreached.
146 McDANIELAND SCHLAGER

moreproblems.Subjectsin the no-delaygroupsweregivena total of 2 hr to


complete the entiretask. The second session for the delay groups began
with the same generalinstructionsas those presentedon the first day. This
was done to reacquaintsubjectswith the procedure.The instructionswere
immediatelyfollowedby one of the two transfersets. None of the subjects
who successfullycompletedthe trainingphase failedto completethe trans-
fer phase.

Resultsand Discussion
Acquisition. For all analyses,the rejectionlevel was set at .05. Mean
solution times and total movesare given in Table3 in parentheses.The to-
tal-movemeasurereflects the sum of the legal moves, illegal moves (i.e.,
movesresultingin states that violate the generalwater-jarproblemrestric-
tions, e.g., overflowinga jar), and restartmoves (a commandto eraseall
previousmovesfrom the screenand startthe problemover).Measurements
of solutiontime werecollectedbecausethe simpleadd and subtractnature
of the movesin this domainmay haveallowedsubjectsto covertlyconsider
and rejectmovesthat werenot enteredinto the computer.Presumablyany
such covertactivitywouldbe reflectedin solutiontime.
Three-way2 (InstructionalLevels)x 4 (TrainingSets) x 4 (Problemsin
Each Set) mixed ANOVAswereperformedon log-normalizedtransforma-
tions of each of the just-mentionedperformancemeasures.Thereweretwo
significanteffects (for each dependentmeasure)involvingthe instructional
variable,both of which were interactions.A two-wayinteractionbetween
instruction and the problem number in each set (i.e., first problem, second
problem, etc.) indicatedthat SMD subjectsrequiredmore time and took
more movesthan MD subjectsprimarilyfor the first problemin each set,
F(3, 354) = 28.95, MSE = .05, and F(3, 354) = 4.20, MSE = .06, respec-
tively. A significant three-wayinteraction(among instruction, problem
number,and set number)furtherindicatedthat this resultwas limited to
Problem Sets 3 and 4, F(9, 1062)= 10.02, MSE= .05, and F(9,
1062)= 2.77, MSE= .06, for time and moves, respectively.Indeedfor the
first problemin ProblemSet 1, SMD subjectswerenominallyfaster and
took slightlyfewermovesthan MD subjects.
Solution times and moves significantlyvariedas a functionof problem
set, F(3, 354) = 565.03, MSE = .07, and F(3, 354) = 163.25, MSE = .09, re-
spectively.This effect was due in part to the fact that the numberof moves
for minimum-pathsolutionvariedacrossproblemsets. The particularpat-
tern of this effect, however,suggeststhat an additionalcomponentwas in-
volved(presumablya learningcomponent).Subjectsspentlongerand took
moremoveson Set 1 problemsthan any otherset, despitethe fact that Set 1
problemscould be solved in just four moves(vs. six moves for Set 2, four
for Set 3, and sevenfor Set 4).
DISCOVERYLEARNING 147

TABLE3
TotalNumberof Movesand SolutionTime for Each Training
Problemas a Functionof Instructionin Experiment2

Moves Timea

ProblemNumber MD SMD MD SMD

Set I

Problem1 13.7 12.4 535.9 490.3


Problem2 20.4 18.0 578.2 444.8
Problem3 8.4 7.4 142.2 114.5
Problem4 7.2 9.2 116.1 165.3
(12.1) (323.4)
Set 2

Problem1 8.5 9.0 128.3 153.8


Problem2 7.1 8.2 77.4 81.6
Problem3 9.0 9.0 92.4 95.8
Problem4 10.0 7.8 86.3 68.8
(8.6) (98.0)
Set 3

Problem1 6.0 9.3 63.0 124.7


Problem2 5.6 4.5 52.4 38.0
Problem3 4.3 4.2 36.8 29.4
Problem4 4.8 4.6 38.5 35.0
(5.4) (52.2)
Set 4

Problem1 9.7 23.3 95.4 309.8


Problem2 8.7 9.2 74.0 71.4
Problem3 9.2 7.7 66.3 45.0
Problem4 8.3 7.3 54.7 39.3
(10.4) (94.4)
Note. Numbersin parenthesesrepresentmean totals for the MD and SMD instruction
groupscombined.
in seconds.
aMeasured

There was also a significant main effect of problem number for solution
time and moves, F(3, 354) = 335.23, MSE = .05, and F(3, 354) = 37.85,
MSE = .06, respectively, such that subjects experienced more difficulty on
the initial problems (in a set) than the later problems, in general. This effect
significantly interacted with problem set for time and moves, indicating
that the difference between initial and later problems was obtained almost
exclusively in Set 1, F(9, 1062) = 39.04, MSE= .05, and F(9, 1062)
= 18.79, MSE = .06, respectively.
These results suggest that, on the first problem set, the MD subjects were
148 McDANIELAND SCHLAGER

searchingfor the correctmoveswith whichto implementthe givenstrategy;


SMD subjects were also searchingfor the correctmoves from which the
strategycould be derived(as indicatedby the pronounceddifficulty dis-
playedby all subjectsin solvingthe first two problemsin Set 1 and the fail-
ure to achieve minimum-pathperformance-four moves-even by the
fourth problem in Set 1). By Problem Set 3, both groups were able to
choose proficientlythe movesneededto implementa strategy(as indicated
by performancequickly asymptotingnear minimum-pathsolutions-four
movesin Set 3 and sevenmovesin Set 4). However,on the first problemof
Sets 3 and 4, respectively,SMD subjectsfirst had to go throughthe extra
effort of discoveringanotherstrategy(as indicatedby these subjects'rela-
tive difficultywith the first problemsin Sets 3 and 4).
We suggestthat this extraeffort to discoveranotherstrategyincludesan
importantprocess.Specifically,the discoveryprocessmay involvethe devel-
opmentor refinementof procedures(possiblyproceduralizations of declar-
atively representedheuristics-see Anderson, 1983; Lewis & Anderson,
1985)for generatingand testingalternativestrategiesor plans. This kind of
knowledge(presumablyacquiredby SMD subjects)should, in turn, prove
useful in transferwhen subjectsattemptto solve a problemgovernedby a
new strategy.This possibilitywas addressedin the analysisof the transfer
data.

Transfer. The mean solutiontimes and the mean numberof moves for
each transferproblemare presentedin Table4. The data werefirst submit-
ted to three-waybetween-subjects analysesof covariance(ANCOVAs),with
instruction(SMD, MD), type of transferproblem(near,remote),and test
interval (immediate,delayed) as the independentvariables. These AN-
COVAswere performedon log-normalizedtransformationsof the total-
move and solution-time measures. Total solution time for Training
Problems1, 5, 9, and 13 was covariedwith the transfer-performance vari-
ables to reducethe possibilitythat any transfereffects would be obscured
becauseof individualdifferencesin problem-solvingskill amongsubjectsin
the experimentalgroups.
Comparedwith remote-transferproblems,near-transferproblemswere
solved in significantly fewer moves (9.0 vs. 19.0), F(1, 228) = 31.99,
MSE = .20, and requiredsignificantlyless time to solve (116.7 sec vs. 282.1
sec), F(1, 228) = 41.09, MSE = .21. This advantagefor near-transferprob-
lems was expected,giventhat the near-transferproblemsweremore similar
to the trainingproblemsthan the remote-transfer problems.The near-trans-
fer problemsalso appearedto be somewhateasierthan the remote-transfer
problems,regardlessof prior training(see Table4 for the control group
means).Therewas a significantmaineffect of instructionfor solutiontime,
with SMD instruction producing faster solutions than MD instruction
(163.8 sec vs. 235.0 sec), F(1, 228) = 7.50, MSE = .21. The SMD instruc-
DISCOVERYLEARNING 149

TABLE4
TotalNumberof Movesand SolutionTimeas a Functionof Type
of TransferProblem,Instruction,and Time of Test in Experiment2

Moves Time"
Transfer
Instruction7Tpe Immediate Delay Mean Immediate Delay Mean

Near
Three-jar
MD 13.5 22.8 18.2 187.5 247.2 217.4
SMD 8.9 14.6 11.8 139.4 220.1 179.8
Control 26.7 941.4
TWo-jar
MD 3.3 2.7 3.0 39.0 34.8 36.9
SMD 2.5 3.3 2.9 30.6 34.8 32.7
Control 4.1 105.3
Remote
Three-jar
MD 24.3 33.2 28.8 378.5 552.0 465.2
SMD 25.1 24.9 25.0 279.6 406.6 343.1
Control 38.0 1115.8
TWo-jar
MD 16.8 13.4 15.1 233.8 207.4 220.6
SMD 8.8 5.4 7.1 94.5 104.4 99.4
Control 4.8 167.5

Note. The test intervalvariableis meaninglessfor the controlgroups.


aMeasuredin seconds.

tions also tended to produce fewer total moves than MD instruction (11.7
vs. 16.3), F(1, 228) = 3.64, p < .06, MSE = .20.
Although the instructional variable did not interact significantly with the
transfer-problem variable, inspection of the means (see Table 4) reveals that
the magnitude of the difference between the instructional groups was
greater for the remote-transferproblems, especially for solution time. These
possible interactions between instruction and problem type may not have
reached statistical significance because of the nonnormal distributions of
scores created by forming conceptually similar groups (near and remote
transfer) from problems that varied dramatically in difficulty. Table 4 gives
an indication of how greatly performance varied between the two problems
in each transfer category; the difference in performance probably resulted
in an inflated error term. This observation motivated separate two-way
(instruction and test delay) between-subjects ANCOVAs for each problem.
These analyses directly test the expectations outlined at the outset of the
experiment.
For both near-transfer problems, as expected, there were no main effects
of instruction or interactions between instruction and delay for either of
the dependent variables (Fs < 1). (Upon inspecting Table 4 with regard to
150 McDANIELAND SCHLAGER

the presentanalysis,the most importantvaluescan be foundin the columns


labeledmean. The valuesassociatedwith the controlconditionscan be ig-
noredfor the moment.)Thesefindingsreplicatethose of Experiment1 in a
second problemdomain, lending further support to the contention that
SMD instructiondoes not enhancetransferrelativeto MD instructionwhen
the transferproblemsare solvedwith the samestrategybut differentmoves
than those used for the trainingproblems.Incidentally,althoughsolution
times were not collected on the river-crossingproblemsin Experiment1,
they weremeasuredby McDaniel(1980). In that study,as in the currentex-
periment,therewereno differencesin solutiontimes.
For both remote-transferproblems,the SMD group displayedsignifi-
cantly faster solution times than did the MD group; F(1, 52) =6.08,
MSE = .06, for the three-jarremote-transfer problem,and F(1, 52) = 6.53,
MSE = .17, for the two-jarremote-transfer problem.As expected,this in-
structional effect did not significantlyinteract with test delay.The instruc-
tional groups did not differ significantlyon the numberof moves,although
therewas a marginallysignificantadvantagefor the SMD group,relativeto
the MD group, for the two-jar remote-transferproblem,F(1, 52) = 3.17,
p < .09, MSE = .22. One other significanteffect emerged;solution times
weresignificantlylonger after delayedtestingthan after immediatetesting
on the three-jarproblem,F(1, 52) = 8.24, MSE = .06.
Finally,performanceof each instructionalgroupwas comparedwith the
performanceof the appropriatecontrolgroup(controlsweregivenno train-
ing). Pairwisecomparisonsusing the Thkey(B) test, with a cutoff signifi-
cance level of .05, revealedreliabledifferenceson each transferproblem
(referto Table4 for the means). On the three-jarnear-transferproblem,
both experimentalgroupsscoredbetterthan the controlgroupon both de-
pendent measures.On the two-jar near-transferproblem, SMD and MD
subjectssolvedthe problemsignificantlyfasterthan did theircontrolcoun-
terparts.On the three-jarremote-transfer problem,the experimentalgroups
again were significantlyfaster at solving the problem than the control
group.These findingssupportthe idea that the trainedsubjectsdeveloped
domain-specificknowledgethat helped them solve certaintypes of water-
jar problems.
The resultsof the two-jarremote-transfer problemindicatethat training
can lead to eithernegativeor positivetransfer.The MD subjectstendedto
make more moves than the control group (p<.08). However,the control
and SMD groupsdid not differ significantlyon eitherdependentmeasure.
In fact, the SMD group had nominallyfaster solution times. This result
suggeststhat SMD learningmay help preventEinstellungeffects (negative
set effects). Inasmuchas the nonoccurrenceof Einstellungcan be an indi-
catorof greaterexpertise(Sweller& Cooper,1985),the currentfindingsup-
ports the idea that SMD trainingis associatedwith developmentof skills
not acquiredby MD subjects.The natureof these skills is addressednext.
DISCOVERYLEARNING 151

Theoretical implications. The key finding was that SMD training was
advantageous, relative to MD training, when the transfer task required use
of a strategy that was different from that used to solve training problems.
As mentioned earlier, one interpretation of this finding is that, during
training, the SMD subjects had to utilize processes for generating and test-
ing alternative strategies, whereas the MD subjects did not. Having to use
these general processes during training apparently practices, refines, and/or
develops them so that they can be used more efficaciously in later novel
problems within the domain. Other research has shown that active hypothe-
sis-testing processes can play a role in learning solution strategies for trans-
formation-type problems (Lewis & Anderson, 1985), and it may be that
differential hypothesis-testing processes (across the instructional groups)
were responsible for the present transfer results.
Alternatively, differential transfer for the SMD versus the MD subjects
might have been due to differential acquisition of particular solution strate-
gies, which may have guided selection of particular move sequences, during
training that were common to the training and transfer problems. At first
glance, it might appear that this common elements explanation would not
be viable in the present context, because the solution strategies for the re-
mote-transfer problems differed explicitly from the strategies governing the
training problems. However, the solution protocols indicated that, during
training, subjects entertained specific erroneous strategies that produced
dead ends. Although these strategies produced dead ends in the training
problem, they still could be useful for the remote-transfer problems. That
is, solving a remote-transfer problem might be accomplished by applying a
strategy that produced a dead end in the training problems. Further, it is
possible that such deadend strategies, common to those required to solve
the remote-transfer problems, were tried more frequently by SMD subjects
during training. If so, enhanced remote transfer by SMD subjects might
have been due to direct application of these common strategies to the trans-
fer problems.
To distinguish between these two explanations, we selected an easily de-
tected strategy that was ineffective in solving the training problems but was
common to the strategy required by the two-jar remote-transfer problem.
The strategy involved simply adding the contents of the medium and small
jars together. We classified subjects who displayed this strategy during
training as "add" subjects, and subjects who did not as "no-add" sub-
jects.4 The first important finding was that both SMD and MD subjects in

'The movedata for each subjectwerecategorizedin termsof incorrectsolutionpaths on


the firstproblemof eachacquisitionset. Wedefinedthis as any sequenceof movesendingin a
returnto the startstateor in the retracingof at least two movesback to a previouslyencoun-
teredstate. Thosesubjectswho at any time attemptedto pursuea pathby addingthe medium
and smalljars only (i.e., the correctstrategyfor solvingthe two-jarremote-transfer
problem)
wereconsideredto be partof the add group.
152 McDANIELAND SCHLAGER

the remote-transfer conditionwereequallylikely to have tried the addition


strategyduring training-18 SMD subjects and 17 MD subjects overtly
tried to add the mediumand smalljars, whereas12 SMD and 13 MD sub-
jects did not try this strategy.
Next we assessedthe degreeto whichexperiencewith the additionstrat-
egy during trainingwould enhance transferto a problem for which the
strategywas required.Wecomparedall subjects(regardlessof instructional
group) comprisingthe add group with all subjectscomprisingthe no-add
groupin termsof performanceon the two-jarremoteproblem.Therewas a
significantadvantagefor the add groupon legal (7.50 vs. 10.35)and restart
moves (.95 vs. 1.82), F(1, 48) = 6.23, MSE= 1.67, and F(1, 48) =4.60,
MSE = .30, respectively,and legal time (75.2 sec vs. 196.1 sec) and restart
time (13.4 sec vs. 30.0 sec), F(1, 48)= 15.05, MSE=.12, and F(1,
48) = 5.79, MSE = 1.19, respectively.5This is an importantfindingbecause,
although others have emphasizedthat initial proceduresthat lead to solu-
tions are not lost (Anderson,Farewell,& Sauers,1984), nothing has been
said of the role that temporarilyabandonedand irrelevantstrategiesplay in
skill acquisition.The presentresultsindicatethat these, too, are retained
and are availablefor use undercertaincircumstances.
Finally,we foundthat the add subjectsdid not performsignificantlybet-
ter on eitherof the near-transferproblems(for the near-transfercondition,
14 of the add subjectswerefrom the SMD groupand 10 werefromthe MD
group; 16 of the no-add subjectswere from the SMD group and 20 were
from the MD group) or the three-jarremote-transferproblem (none of
these problemsrequireduse of the add strategy).
Thus, an ineffectivestrategyexploredin trainingfacilitatedtransferonly
for a transferproblemin which the strategywas required.This finding is
consistentwith the idea that transferperformancewas determined,in part,
by specificsolution strategiesthat werecommonto the trainingand trans-
fer problems.This mechanism,however,probablydid not account for the
instructional differences in remote transfer, because the instructional
groupsdid not differ in their explorationof at least one possible addition
strategy(i.e., the add strategy).Moreover,the transferdifferencesassoci-
ated with varyingdegreesof explorationof the add strategyincludedsignif-
icant differencesin numberof moves,whereasthe significantinstructional
effects werelimited to solution times. Thus, it seems that the subjectsin
both instructionalgroupsinitially attemptedto use their learnedstrategy,

5Forall analysesreportedin Experiment2, parallelanalyseswereconductedfor each type


of move (legal,illegal, and restart)and for time to completeeach type of move. Becausethe
resultsof theseanalyseswerefor the most partredundantwith the total moveand total time
analyses,for purposesof clarityand brevity,we omittedreportingthesedetailedanalyses.For
the analysiscomparingthe post hoc add and no-addgroups,however,only the detailedanaly-
ses wereperformed.
DISCOVERYLEARNING 153

failed, and began to look for new strategies. The SMD subjects were faster
than MD subjects at developing or generating alternative strategies.

GENERAL DISCUSSION

The present analysis of the effects of discovery learning on transfer per-


formance refines and extends previous views (see Shulman & Keislar, 1966)
that posited seemingly straightforward functional relationships between
discovery in acquisition and the efficacy of transfer. In our view, discovery
learning does not produce better skill at applying the discovered strategy
during transfer. This claim was supported by findings in two problem do-
mains (river-crossing problems and water-jar problems). In both problem
domains, having to discover a general strategy versus being given a strategy
did not enhance transfer of the strategy, even when transfer occurred several
days after acquisition and required a new implementation of the general
strategy (i.e., a new move combination).
The benefit of discovering a strategy seems to be that it encourages the
development, practice, and/or refinement of procedures that aid the learner
in generating searches for new strategies. This was evidenced in Experiment
2 by the finding that having to discover a strategy (i.e., the SMD condition)
enhanced transfer relative to being given the strategy (i.e., the MD condi-
tion) when the transfer problem required a new strategy.
It is not yet apparent whether the present pattern of effects will gen-
eralize to domains other than transformation problems. Clearly, other
problem domains require different kinds of skills and knowledge. For in-
stance, analogy and series-extrapolation problems require skill in appre-
hending relations among problem elements (Greeno, 1978), whereas
problems of estimating average speed require knowledge about principles
and knowledge about specific symbols and their referents (Reed & Saave-
dra, 1986). In principle, however, the idea that discovery learning is benefi-
cial primarily when transfer requires discovery of new information (e.g.,
strategies) that is similar in kind to that discovered in acquisition (e.g.,
other strategies) is potentially general in nature. Indeed, this idea fares well
in revisiting the seemingly equivocal body of literature comparing discov-
ery and expository learning. Examining the literature from the just-men-
tioned perspective suggests the following uniformity: Discovery instruction
significantly benefits transfer relative to expository instruction only when
discovery instruction requires learners to discover information during ac-
quisition that is similar to the kind of information that has to be discovered
to complete the particular transfer task and when expository instruction
does not require discovery of such information in acquisition (e.g., see
the results of Corman, 1957; Guthrie, 1967; Katona, 1940; Scandura, 1964;
Waters, 1928).
154 McDANIELAND SCHLAGER

An important question that remains is how discovery of certain informa-


tion in acquisition facilitates discovery of similar information in transfer.
Although our data do not directly address this issue, one possibility is that,
in the course of discovery, the learner acquires heuristic procedures for dis-
covering a particular kind of knowledge about the domain, and these pro-
cedures are then applied in the context of the transfer problem. Note that
we are not suggesting that learners necessarily acquire general learning-to-
learn procedures. For the present studies, we suggest that learners acquired
domain-specific procedures.
This idea is consistent with Kolers's work on learning and transfer (Ko-
lers & Magee, 1978; Kolers & Roediger, 1984), in which learning by doing
(e.g., learning to read inverted text) is characterized by acquisition of spe-
cific analytic procedures that can be applied to improve performance if and
only if the test task requires use of the specific analytic procedures acquired
in training. Our findings also parallel the "transfer-appropriateprocessing"
effects reported in the literature on memory (e.g., McDaniel, 1981;
McDaniel, Friedman, & Bourne, 1978; Morris, Bransford, & Franks, 1977).
These studies demonstrated that memory is best when the type of informa-
tion emphasized by the encoding activity is similar to the type of informa-
tion required by the memory test. Thus, the present results, taken in
conjunction with a wide array of work in human learning (e.g., work on
reading skills, Kolers & Magee, 1978; word list memory, McDaniel et al.,
1978; sentence memory, McDaniel, 1981; implicit memory, Roediger,
Weldon, & Challis, 1989; and training on verbal learning tasks, Postman &
Schwartz, 1964), imply that a "transfer-appropriate training" principle
may be a pervasive and general feature of human learning.

Educational Implications
We conclude with some reflections on the educational implications of the
present research. Clearly, the increased effort and training time associated
with discovery modes of instruction will not always be beneficial. Specifi-
cally, discovery learning per se does not seem to provide subjects with more
insight into the information targeted for learning (e.g., a particular strat-
egy). Once learned, the information (e.g., a strategy) seems to be transfer-
red with equal facility, regardless of whether or not it was discovered or
explicated for the learner. More particularly, having to discover problem-
relevant information (e.g., a domain-specific strategy) does not appear to
produce better generalizability of that strategy to other problems in the do-
main for which the strategy can be applied (see also, Guthrie, 1967; Norton,
Graham, & Merrill, 1977). And discovery learning per se does not seem to
enable the learner to identify more quickly the one useful strategy from a
repertoire of several different strategies acquired in training (e.g., near-
transfer performance in Experiment 2).
DISCOVERYLEARNING 155

Discovery learning does appear useful under certain conditions, however.


Specifically, transfer will be enhanced if the transfer task necessitates dis-
covery of the type of information that had to be discovered in acquisition.
To be efficacious, instructional modes that utilize discovery learning appar-
ently need to be designed so that the type of information to be discovered
in acquisition is also the type of information for which discovery is required
in transfer. Also, it appeared that discovery learning could potentially con-
fer an advantage via exploration of dead ends during training that become
useful in the transfer tasks. This advantage, however, would be manifested
only to the extent that alternative instructional conditions do not provide
experience with the same dead ends.
Clearly, more work is warranted before these prescriptions can be fully
adopted for pedagogical practice. One implicit assumption of a discovery-
learning orientation is that requisite knowledge can be discovered by the
learner. Although the majority of subjects were able to do this in the
present problem domains, discovery may not always be possible (or may be
much too inefficient) for knowledge underlying performance in more com-
plex domains (e.g., see Reed & Saavedra, 1986). Moreover, most domains
of knowledge take semesters or years to acquire and involve learning a com-
plex knowledge structure with many interrelated levels of abstract and spe-
cific knowledge. The present research probably most closely parallels
instructional situations in which one problem-solving strategy or abstrac-
tion is assimilated in one lesson rather than one in which a body of complex
concepts is learned over the course of many lessons and must be interre-
lated and integrated into extensive knowledge structures.
Even in this circumscribed instructional situation, for educationally ori-
ented problem domains (e.g., physics and mathematics), the components
of problem-solving expertise may not map neatly onto the kind of knowl-
edge components identified for the transformational problems used herein
(i.e., strategies and move sequences). For example, in an evaluation of in-
structional methods for improving students' estimates of one kind of physi-
cal functional relation (average speed), Reed and Saavedra (1986) identified
three critical knowledge components: specific facts, data trends, and princi-
ples. The extent to which discovery and transfer of these kinds of knowl-
edge components would parallel that for the knowledge components that
we have considered is unknown.
The present study demonstrates the value of analyzing discovery learning
in terms of more precise components of knowledge that must be discovered
during learning and in terms of the distinction between applying that par-
ticular knowledge in transfer situations (for which discovery may not con-
fer an advantage) versus having to generate or discover new knowledge for
successful transfer (for which discovery may confer an advantage). It is this
general approach that may provide a useful foundation for future work on
156 McDANIELAND SCHLAGER

discoverylearningin ecologicallyvalid problem-solvingand instructional


domains.

ACKNOWLEDGMENTS

Preparationof this articlewas supportedin part by the National Institute


of Child Health and Human DevelopmentGrant HD 23984 to M. A.
McDaniel.
Experiment2 is based on M. S. Schlager'smaster'sthesis, submittedto
the Departmentof Psychologyat the Universityof Colorado.
We are indebtedto Peter Polson for his guidanceand supportthrough
this phaseof the researchand for his commentson an earlierversionof this
article.We also thankJeromeBusemeyer,LaurenResnick,and two anony-
mous reviewersfor their helpful commentson an earlierdraft. Finally,we
appreciateTomDixon'sefforts in collectingthe data for Experiment1.

REFERENCES

Anderson,J. R. (1983). The architectureof cognition.Cambridge,MA: HarvardUniversity


Press.
Anderson,J. R., Farewell,R., & Sauers,R. (1984).Learningto programin LISP.Cognitive
Science,8, 87-129.
Ausubel,D. P. (1964).Some psychologicaland educationalimitationsof learningby discov-
ery.ArithmeticTeacher,11, 290-302.
Bassok,M., & Holyoak,K. J. (1989).Interdomaintransferbetweenisomorphictopicsin alge-
bra and physics.Journalof ExperimentalPsychology:Learning,Memory,and Cognition,
15, 153-166.
Brown,A. L., & Kane,M. J. (1988). Preschoolchildrencan learn to transfer:Learningto
learnand learningfromexample.CognitivePsychology,20, 493-523.
Bruner,J. S. (1961).The act of discovery.HarvardEducationalReview,31, 21-32.
Catrambone,R., & Holyoak, K. J. (1989). Overcomingcontextuallimitationson problem-
solvingtransfer.Journalof ExperimentalPsychology:Learning,Memory,and Cognition,
15, 1147-1156.
Corman,B. R. (1957).The effect of varyingamountsand kindsof informationas guidancein
problemsolving.PsychologicalMonographs,71 (WholeNo. 431).
Friedlander,B. Z. (1965).A psychologist'ssecondthoughtson concepts,curiosity,and discov-
ery in teachingand learning.HarvardEducationalReview,35, 18-38.
Greeno,J. G. (1978).Naturesof problem-solving abilities.In W. K. Estes(Ed.), Handbookof
learningand cognitiveprocesses:Human informationprocessing(Vol. 5, pp. 239-270).
Hillsdale,NJ: LawrenceErlbaumAssociates,Inc.
Guthrie,J. T. (1967). Expositoryinstructionversusa discoverymethod.Journalof Educa-
tionalPsychology,58, 45-49.
Jeffries,R. M. (1979).The acquisitionof expertiseon missionaries-Cannibalsand waterjug
problems(Doctoraldissertation,Universityof Coloradoat Boulder,1978).DissertationAb-
stractsInternational,39, 5614B.
DISCOVERYLEARNING 157

Katona,G. (1940).Organizingand memorizing.New York:ColumbiaUniversityPress.


Kolers,P. A., & Magee,L. E. (1978).Specificityof pattern-analyzing skillsin reading.Cana-
dianJournalof Psychology,32, 43-51.
Kolers,P. A., & Roediger,H. L., III. (1984).Proceduresof mind.Journalof VerbalLearning
and VerbalBehavior,23, 425-449.
Lewis,M. W., & Anderson,J. R. (1985). Discriminationof operatorschematain problem
solving:Learningfromexamples.CognitivePsychology,17, 16-65.
Luchins,A. S. (1942). Mechanizationin problemsolving.PsychologicalMonographs,54 (6,
SerialNo. 248).
McDaniel,M. A. (1980). Bottom-upand top-downacquisitionof expertiseon rivercrossing
problems:A studyof knowledgeacquisition(Doctoraldissertation,Universityof Colorado
at Boulder,1980).DissertationAbstractsInternational,41, 1548B.
McDaniel,M. A. (1981).Syntacticcomplexityand elaborativeprocessing.Memory& Cogni-
tion, 9, 487-495.
McDaniel,M. A., Friedman,A., & Bourne,L. E., Jr. (1978).Remembering the levelsof in-
formationin words.Memory& Cognition,6, 156-164.
Morris,C. D., Bransford,J. D., & Franks,J. J. (1977).Levelsof processingversustransferap-
propriateprocessing.Journalof VerbalLearningand VerbalBehavior,16, 519-533.
Norton, R. F., Graham,S. L., & Merrill,M. D. (1977,April). Theeffectsofpractice-itemva-
riety,practicestrategyand trainingmadein a rule-findingtask. Paperpresentedat the an-
nual meetingof the AmericanEducationalResearchAssociation,New York.
Novick,L. R. (1988).Analogicaltransfer,problemsimilarity,and expertise.Journalof Exper-
imentalPsychology:Learning,Memory,and Cognition,14, 510-520.
Polson, P. G. (1988).The consequencesof consistentand inconsistentuser interfaces.In R.
Guindon(Ed.), Cognitivescienceand its applicationsfor human-computerinteraction(pp.
59-108). Hillsdale,NJ: LawrenceErlbaumAssociates,Inc.
Postman,L., & Schwartz,M. (1964).Studiesof learningto learn:I. Transferas a functionof
methodof practiceand classof verbalmaterials.Journalof VerbalLearningand VerbalBe-
havior,3, 37-49.
Reed,S. K., Ernst,G. W., & Banerji,R. (1974).The role of analogyin transferbetweensimi-
lar problemstates.CognitivePsychology,6, 436-450.
Reed, S. K., & Saavedra,N. C. (1986). A comparisonof computation,discovery,and graph
proceduresfor improvingstudents'conceptionof averagespeed.Cognitionand Instruction,
3, 31-62.
Roediger,H. L., Weldon,M. S., & Challis,B. H. (1989).Explainingdissociationsbetweenim-
plicitand explicitmeasuresof retention:A processingaccount.In H. L. Roediger& F. I. M.
Craik(Eds.), Varietiesof memoryand consciousness:Essays in honour of Endel Tulving
(pp. 3-41). Hillsdale,NJ: LawrenceErlbaumAssociates,Inc.
Scandura,J. M. (1964).An analysisof expositoryand discoverymodesof problemsolvingin-
struction.TheJournalof ExperimentalEducation,33, 149-159.
Schlager,M. S. (1984). Top-downversusbottom-uplearning:A study of knowledgeacquisi-
tion and transfer.Unpublishedmaster'sthesis,Universityof Coloradoat Boulder.
Shulman,K. S., & Keislar,E. R. (Eds.). (1966).Learningby discovery:A criticalappraisal.
Chicago:RandMcNally.
Stein, B. S., Way,K. R., Benningfield,S. E., & Hedgecough,C. A. (1986). Constraintson
spontaneoustransferin problem-solving tasks.Memory& Cognition,14, 432-441.
Sweller,J., & Cooper,G. A. (1985).The use of workedexamplesas a substitutefor problem
solvingin learningalgebra.Cognitionand Instruction,2, 59-89.
Thorndike,E. L., & Woodworth,R. S. (1901).The influenceof the improvement in one men-
tal functionupon the efficiencyof otherfunctions.PsychologicalReview,8, 247-261, 384-
395, 553-564.
158 McDANIELAND SCHLAGER

Waters,R. H. (1928).The influenceof tuition upon ideationallearning.Journalof General


Psychology,1, 534-547.
Weisberg,R. W. (1988).Problemsolvingand creativity.In R. J. Sternberg(Ed.), Thenature
of creativity(pp. 148-176).New York:CambridgeUniversityPress.

APPENDIX A

This appendixcontainsdescriptionsof the threegeneralsolutioncategories


used for classifyingsubjects'move protocols from the transferproblems.
The classificationprocesswas formalizedthroughan algorithmconsisting
of the followingsteps. If a perfectmatchwas foundbetweena subject'sso-
lution and one of the threestrategytemplates,the solutionwas classifiedas
an instanceof that strategy.If no matchwas found, the nearestmatchwas
taken, providedthat the subject'smoves wereconsistentwith 80%0of the
solution as prescribedby the template. If a tie existed, the classification
that resultedin the smalleramountof deviationsfrom the templatewas se-
lected. In other words, a single two-movedetour was rated better than 2
one-movedetours.A subject'ssolution was scoredas "unclassifiable"if it
deviatedby morethan 20% from each of the templates.

Orcs-Switch-Orcs
This strategycan be describedas:
Move wives either for four moves or until a boatload (or nearlyfull
boatload) of husbands can be brought across, move husbands to
achievea balance,returna husband-wifepair,movethe remainderof
the husbands,movethe rest of the orcs across.

Balance
This "paired"strategymerelyinvolvestransportingtwo pairs of husbands
and wives acrossthe river,returninga pair, and continuingin like fashion
until the problemis solved.This will not producea minimum-pathsolution
for the transferproblemused.

Orcs-Switch-Orcs-Balance
This "mixed"strategyis a hybridof the OSO and the balanceapproaches.
The movesrepresentingthis strategyare like those for OSO until the inter-
change,then husband-wifepairs are moved(as in the balancestrategy)to
finish the problem;or husband-wifepairsare moveduntil the interchange,
then an OSO sequenceis used to finish the problem.
DISCOVERYLEARNING 159

APPENDIXB

Experiment2: ExperimentalTaskSet
JarA JarB Jar C Goal Formula

ProblemSet

Set 1 123 24 10 85 1A - 2B + 1C
33 89 34 54 1B - 2C + 1A
73 20 21 51 1A-2C+ 1B
7 36 111 46 1C-2B+ 1A
Set 2 3 108 23 33 1B - 3C - 2A
136 32 3 34 1A- 3B - 2C
21 11 122 37 1C-3A-2B
15 158 11 91 1B- 3A- 2C
Set 3 7 24 98 60 1C - 1B - 2A
78 10 21 37 1A- 1C- 2B
93 40 3 47 1A - 1B - 2C
28 138 3 104 1B - 1A- 2C
Set 4 97 23 4 67 1A - 2B + 4C
43 91 18 77 1B-2A + 4C
13 46 174 134 1C- 2B + 4A
129 3 35 71 1A - 2C + 4B

7ypeof TansferProblem

Three-jar
Near 117 22 18 59 1A - 1B - 2C
Remote 15 48 166 124 1C - 6A + 1B
Two-jar
Near 67 32 129 97 1C- 1B
Remote 34 117 8 42 1A + 1C

You might also like