Allan Collins

CHAPTER 4

Cognitive Apprenticeship

,,"I.~

n

,).u­,I,,mm-

;tto

en­lVi-

;;1-

t tIle

~;tl'r.

~ical

r('t~.

Throughout most of history, teaching andlearning have been based on apprentice­ship. Children learned how to speak, growcrops, construct furniture, and make clothes.But they didn't go to school to learn thesethings; instead, adults in their family and intheir communities showed them how, andhelped them do it. Even in modern societies,we learn some important things throughapprenticeship: we learn our first languagefrom our families, employees learn criticaljob skills in the first months ofa new job, andscientists learn how to conduct world-classresearch by working side-by-side with seniorSCientists as part of their doctoral training.But for most other kinds of knowledge,schooling has replaced apprenticeship. Thenumber of students pursuing an educationhas dramatically increased in the last twocenturies, and it gradually became impossi­ble to use apprenticeship on the large scaleof modern schools. Apprenticeship requiresa very small teacher-ta-learner ratio, and thisis not realistic in the large educational sys­tems of modem industrial economies.

Even in modern SOCieties, when some­one has the resources and a strong desireto learn, they often hire a coach or tutorto teach them by apprenticeship - demon·strating that apprenticeship continues tobe more effective even in modern soci­eties. If there were some way to tap intothe power of apprenticeship, without incur­ring the large costs associated with hiringa teacher for every two or three students,it could be a powerful way to improveschools. In the 1970S and 19805, I was doingresearch at the intersection of educationand new computer technology, and alongwith many other scholars, J was studyinghow this new tec~nology could help usto transform schooling. Working with mycolleague John Seely Brown, we began tobelieve that we could develop sophisticatedcomputer-based learning environments thatcould provide students with apprenticeship­like experiences, providing the type ofdose attention and immediate responsethat has always been associated withapprenticeship.

47

.8 TllR (;MIRRI DCE !IANDBOOK Of' TilE LEARNING SCIENCES

From Traditional to CognitiveApprenticeship

In her study of a tailor shop in Africa,Lave identified the central features of tra­ditional apprenticeship (Lave, 1I;J88). First,traditional apprenticeship focuses closely onthe specific methods for carrying out tasksin a domain. Second, skills are instrumentalto the accomplishment of meani(lgful real­world tasks, and learning is em~ddcd in asocial and functional context, unlike school­ing, where skills and knowledge are usu­ally ahstratted from their use in the world.Apprentices learn domain.sped6c methodsthrough a combination of what Lave calledobservation, coaching, and practice. In thissequence of activities, the apprentice repeat­edly observes the master and his or herassistants executing (or modeling) the tar­get process, which usually involves a num­ber of different, but interrelated subskills.The apprentice then attempts to execute theprocess with guidance and help from themaster (i.e., coaching). A key aspect ofcoaching is guided participation: the closeresponsive support which the master pro­vides to help the novice complete an entiretask, even before the novice has acquiredevery skill required. As the learner mas­ters increasing numbers of the componentskills, the master reduces his or her partici­pation, providing fewer hints and less feed­back to the learner. Eventually, the masterfades away completely, when the appren­tice has learned to smoothly execute thewhole task.

Of course, most of us think of verytraditional trades when we hear the term"apprenticeship" -like shoemaking or farm­ing. John Seely Brown and I realized that theconcept ofapprenticeship had to be updatedto make it relevant to modern subjectslike reading, writing, and mathematics. WecaUed this updated concept of apprentice­ship "cognitive apprenticeship~ to empha­size two issues (Brown, Collins, & Duguid,1989; Collins, Brown, & Newman, 1989).

First, the term "apprenticeship" empha­sized that cognitive apprenticeship wasaimed primarily at teaching processes thatexperts use to handle complex tasks. Like

traditional apprenticeship, cognitive appren­ticeship emphasizes that knowledge must beused in solving real-world problems. Con­ceptual knowledge and factual knowledgearc learned by being used in a variety ofcontexts, encouraging both a deeper under­standing of the meaning of the concepts andfacts themselves, and a rich web of mem­orable associations between them and theproblem solving contexts. This dual focuson expert processes and learning in contextare shared by both traditional apprenticeshipand cognitive apprenticeship.

Second, "cognitive" emphasizes that thefocus is on cognitive skills and processes,rather than physical ones. Traditional ap­prenticeship evolved to teach domains inwhich the process of carrying out targetskills is externally visible, and thus readilyavailable to both student and teacher forobservation, comment, refinement, and cor­rection, and the process bears a relativelytransparent relationship to concrete prod­ucts. But given the way that most subjectsare taught and learned in school, teacherscannot make fine adjustments in students'application of skill and knowledge to prob­lems and tasks, because they can't see thecognitive processes that are going on in stu­dents' heads. By the same token, studentsdo not usually have access to the cogni­tive problem solving processes of instruc­tors as a basis for learning through obser­vation and mimicry. Before apprenticeshipmethods can be applied to learn cognitiveskills, the learning environment has to bechanged to make these internal thought pro­cesses externally visible. Cognitive appren­ticeship is designed to bring these cognitiveprocesses into the open, where students canobserve, enact, and practice them.

'[nere are two major differences betweencognitive apprenticeship and traditional ap­prenticeship. First, because traditional ap­prenticeship is set in the workplace, theproblems and tasks that are given to learn­ers arise not from pedagogical concerns, butfrom the demands ofthe workplace. Becausethe job selects the tasks for students to prac­tice, traditional apprenticeship is limited inwhat it can teach. Cognitive apprenticeshipdiffers from traditional apprenticeship in

that the taskl:illustrate theand methods,applying thesand to increslowly, so th~els can be iIsequenced toof learning.

Second, .....ship emphasitext of theiremphasizes fit can be uS'Cognitive aFto diverse se·man principJto apply thei

A FramewoApprentice

Cognitive 3Jdimensionsenvironmen1

and sociololCollins, Hav

Contenl.

Recent cogrferentiate tlfor cxpcrtis(ter the exp:dures assocwhat reseaDomain knfacts, and 1with a parthe type cfound insddemonstratedge in reaphonics rul

Domainsufficient fvidcs insufabout howplish tasksrecently btkJwwledge

COGNITIVE APPRENTICESHIP

,,t

e..n.pin

that the tasks and problems are chosen toillustrate the power of certain techniquesand methods, to give students practice inapplying these methods in diverse settings,and to increase the complexity of tasksslowly, so that component skills and mod­els can be integrated. In short, tasks arcsequenced to reflect the changing demandsof learning.

Second, whereas traditional apprentice­ship emphasizes tcaching skills in the con­text of their use, cognitive apprenticeshipemphasizes generalizing knowledge so thatit can be used in many different settings.Cognitive apprenticeship extends practiceto diverse settings and articulates the com­mon principles, so that students learn howto apply their skills in varied contexts.

A Framework for CognitiveApprenti.ceship

Cognitive apprenticeship focuses on fourdimensions that constitute any learningenvironment: content, method, sequencing,and sociology (see Table 4.1, taken fromCollins, Hawkins, & Carver, 1991).

Content

Recent cognitive research has begun to dif­ferentiate the types of knowledge requiredfor expertise. Ofcourse, experts have to mas­ter the explicit concepts, facts, and proce­dures associated with a specialized area ­what researchers caJl domain knowledge.Domain knowledge includes the concepts,facts, and procedures explicitly identifiedwith a particular subject matter. This isthe type of knowledge that is generallyfound in school textbooks, class lectures, anddemonstrations. Examplcsofdomain knowl­edge in reading are vocabulary, syntax, andphonics rules.

Domain knowledge is necessary but notSuffiCient for expert performance. It pro·vides insufficient clues for many studentsabout how to solve problems and accom­plish tasks in a domain. Psychologists haverecently been trying to explicate the tacitknowledge that Supports people's ability to

___L_

49

make use of these concepts, facts, and proce­dures to solve real-world problems (also seeBransford et aI., this volume). I call this sec­ond kind of knowledge strategic knowledge.Research has identified three kinds of strate­gic knowledge:

J. Heuristic strategies arc generally effectivetechniques and approaches for accom­plishing tasks that might be regardedas "tricks of the trade"; they don'talways work, but when they do, they arequite helpful. Most heuristics are tacitlyacquired by experts through the practiceof solving problems. However, there havebeen noteworthy attempts to addressheuristic learning explicitly (Schoenfeld,1(;185). In mathematics, a heuristic forsolVing problems is to try to find a solu­tion for simple cases and sec if the solu­tion generalizes.

::. Contml strategies, or metacognitilJe strate­gies, control the process of carrying out atask. Control strategies have monitoring,diagnostic, and remedial components;decisions about how to proceed in atask generally depend on an assessmentof one's current state relative to one'sgoals, on an analysis ofcurrent difficulties,and on the strategies available for dealingwith difficulties. For example, a compre·hension monitoring strategy might be totry to state the main point of a sectiononc has just read; if one cannot do so,then it might be best to reread parts ofthe text.

j. Learning strategies are strategies for learn­ing domain knowledge, heuristic strate·gies, and control strategies. Knowledgeabout how to learfl\ ranges from generalstrategies for explor'ing a new domain tomore specific strategies for extending orreconfiguring knowledge in solving prob­lems or carrying out complex tasks. Forexample, if students want to learn to solveproblems better, they need to learn howto relate each step in the example prob­lems worked in textbooks to the prin·ciples discussed in the text (Chi, et al.,1989). If students want to write better,they need to learn to analyze others' textsfor strengths and weaknesses.

Table 4.1. Principles for Designing Cognitive Apprenticeship Environments

Content Types of knowledge required for expertiseDomain knowledge subject matter sp<"<:in.c concepts, facts, and proceduresHeuristic strategies generally applicable techniques for accomplishing tasksControl strategies Il:eneral approaches for directing one's solution processLearning strategies knowledge about how to learn new concepts, facts, and

procedures

Method Ways to promOte the development of expertiseModeling teacher }X'rforms a task so students can observeCoaching. teacher observes and facilitates while students perform a

,,' taskScaffolding teacher provides supports to help the student perform a taskArticulation teacher encourages students to verbalize their knowlooge

and thinkingReflection teacher enables students to compare their performance

with othersExploration teacher invites students to pose and solve their own

problems

Sequelldllg Keys to ordering learning activitiesIncreasing complexity meaningful tasks gradually increasing in difficultyIncreasing diversity practice in a variety of situations to emphasize broad

applicationGlobal to local skills focus on conceptualizing the whole task before executing

the partsSociology Social characteristics of learning environments

Situated learning students learn in the context of working on realistic tasksCommunity of practice communication about different ways to accomplish

meaningful tasksIntrinsic motivation students set personal goals to seek skills and solutionsCooperation students work together to accomplish their goals

THE CAMBRIDGE IIANDBOOK OF TH£ I.F.....RNING SCIF.NCES;0

Method

Teaching methods that emphasize appren·ticeship give students the opportunity toobserve, engage in, and invent or discoverexpert strategies in context. The six teachingmethods associated with cognitive appren­ticeship faU roughly into three groups. Thefirst three methods (modeling, coaching,and scaffolding) are the core of traditionalapprenticeship. They are designed to helpstudents acquire an integrated set of skillsthrough processes ofobservation and guidedpractice. The next t\Vo methods (articula­tion and reflection) are methods designedto help students to focus their observationsof expert problem solving and to gain con­scious access to (and control of) their ownproblem solving strategies. The final method(exploration) is aimed at encouraging learner

autonomy, not only in carrying out expertproblem solving processes but also in defin­ing or formulating the problems to besolved.

I. Modeling involves an expert performing atask so that the stude-nts can observe andbuild a conceptual model of the processesthat are required to accomplish it. In cog­nitive domains, this requires the exter­nalization of usually internal processesand activities. For example, a teachermight model the reading process by read~

ing aloud in one voice, while verbalizingher thought processes in another voice(ColJins & Smith, illS:!). In mathematics,Schoenfeld (1985) models the process ofsolving problems by having students bringdifficult new problems for him to solvein class.

4

5

'-' l~~ !u; J'J' .:.;., .~.....'" ........_...... _...;,.

COGNITIVE APPRENT1Cl::SIIJP5'

2. Coaching consists of observing studentswhile they carry out a task and offer­ing hints, challenges, scaffolding, feed­back, modeling, reminders, and new tasksaimed at bringing their performancecloser to expert performance. Coaching isrelated to specific events or problems thatarise as the student attempts to accom­plish the task. In Palincsar and Brown's(1984) reciprocal teaching of reading, theteacher coaches students while they askquestions, clarify their difficulties, gener­ate summaries, and make predictions.

3· Scaffolding refers to the supports theteacher provides to help the student carryout the task. Coaching refers broadly toall the different ways that coaches fos­ter learning, whereas scaffolding refersmore narrowly to the supports providedto the learner. These supports can takeeither the form of suggestions or help, asin Palincsar and Brown's (11)84) recipro­cal teaching, or they can take the formof physical supports, as with the cuecards used by Scardamalia, Bereiter, andSteinbach (1984) to facilitate writing, orthe short skis used to teach downhil1 ski­ing (Burton, Brown, & Fischer, 1984). Fad­ing involves the gradual removal of sup­ports until students are on their own.

4· Articulation includes any method of get­ting students to explicitly state theirknowledge, reasoning, or problem solv­ing proceSSes in a domain. Inquiry teach­ing (Collins & Stevens, 198~J is a strat­egy of questioning students to lead themto articulate and refine their understand­ing. Also, teachers can encourage studentsto articulate their thoughts as they carryout their problem solving, or have stu­dents assume the critic or monitor rolein cooperative activities in order to artic­ulate their ideas to other students. Forexample, an inquiry teacher in readingmight question students about why onesummary of the text is good hut anotheris poor, in order to get them to formulatean explicit model of a good summary.

5· Reflection involves enabling students tocompare their own problem solving

processes with those ofan expert, anotherstudent, and ultimately, an internal cog­nitive model of expertise. Reflection isenhanced by the use ofvarious techniquesfor reproducing or "replaying" the per­formances of both expert and novice forcomparison. Some form of "abstractedreplay," in which the critical features ofexpert and student performance are high­lighted, is desirable (Collins & Brown,1988). For reading or writing, methodsto encourage reflection might consist ofrecording students as they think out loudand then replaying the tape for compari­son with the thinking ofexperts and otherstudents.

6. ExpLoration involves guiding students to amode of problem solving on their own.Enabling them to do exploration is crit­ical, if they are to learn how to framequestions or problems that are interest­ing and that they can solve. Explorationas a method of teaching involves set­ting general goals for students and thenencouraging them to focus on particu­lar subgoals of interest to them, or evento revise the general goals as they comeupon something more interesting to pur~

sue. For example, the teacher might sendthe students to the library to investigateand write about theories as to why thedinosaurs disappeared.

Sequencing

Cognitive apprenticeship provides someprinciples to guide the sequencing of learn­ing activities.

1. Increasing complexity refers to the con­struction of a sequence of tasks such thatmore and more of the skills and con­cepts necessary for expert performanceare required (Burton, Brown, & Fischer,H)8-+; White, 1984). For example, in read­ing increasing task complexity might con­sist of progressing from relatively shorttexts, with simple syntax and concretedescription, to texts in which complexlyinterrelated ideas and the use of abstrac­tions make interpretation more difficult.

5'

-----------..TIl I:': CAMRRIDClo; BANDIIOOK OF TIlE I.EARNING SCIENU;S

,.

2. Increasing diversity refers to the construc­tion of OJ sequence of tasks in which awider and wider variety of strategies orskills are required. As a skill becomes welllearned, it becomes increasingly impor­tant that tasks requiring a diversity ofskills and strategies be introduced so thatthe student learns to distinguish the con­ditions under which they do (and donot) apply. MQreover, 3S students learn

•to apply skills to more diverse problems,their strategies acquire a richer net ofcon­textual associations and thus are morereadily available for usc with unfamil­iar or novel problems. For mathematics,task diversity might be attained by inter­mixing very different types of problems,such as asking students to solve problemsthat require them to use a combinationof algebraic and geometric concepts andtechniques.

3. Global before local Skills. In tailoring (Lave,1988) apprentices learn to put together agarment from precut pieces before learn­ing to cut out the pieces themselves. Thechief effect of this sequencing principleis to allow students to build a conceptualmap before attending to the details of theterrain (Norman, 1973). Having a dearconceptual model of the o\'crall activityhelps learners make sense of the portionthat they are carrying out, thus improvingtheir ability to monitor their own progressand to develop attendant self~correction

skills. In algebra, for example. comput­ers might carry out low-level computa­tions - the local skills - so that studentscan concentrate on the global structureof the task, and the higher order reason­ing and strategies required to solve a com­plex, authentic problem.

Sociology

Tailoring apprentices learn their craft notin a special, segregated learning environ­ment, hut in a busy tailoring shop. Theyare surroundf'd both by masters and otherapprentices, all engaged in the target skillsat varying levels of expertise. And they are

expected, from the beginning, to engagein activities that contribute directly to theproduction of actual garments, advancingquickly toward independent skilled produc~

tion. As a result, apprentices learn skills inthe context of their application to real-worldproblems, within a culture focused on anddefined by expert practice. Furthermore, cer~tain aspects of the social organization ofapprenticeship encourage producth"e beliefsabout the nature of learning and of exper­tise that are significant to learners' moti­vation, confidence, and most importantly,their orientation toward problems that theyencounter as they learn. These considera­tions suggest several characteristics affectingthe sociology of learning.

I. Situated learning. Acritical eJement in fos­tering learning is having students carryout tasks and solve problems in an envi­ronment that reflects the nature of suchtasks in the world (Brown, Collins, &Duguid, H)Sq; Lave & Wenger, 19(1). Forexample, reading and writing instructionmight be situated in the context of stu·dents putting together a book on whatthey learn about science. Dewey cre­ated a situated learning environment inhis experimental school by having thestudents design and build a clubhouse(Cuban, 1984), a task that emphasizesarithmetic and planning skills.

2;. Community of practice refers to the cre­ation of a learning environment in whichthe participants actively communicateahout and engage in the skills involved inexpertise (Lave & Wenger, '991; Wenger,19(8). Such a community leads to a senseof ownership, characterized by personalinvestment and mutual dependency. Itcannot be forced, but it can be fosteredby common projects and shared experi­ences. Activities designed to engender acommunity of practice for reading mightengage students in discussing how theyinterpret particularly difficult texts.

'). Intrinsic motivatioll. Related to the issueof situated learning and the creation ofa community of practice is the need topromote intrinsic motivation for learning.

,~.. , ~'._.-.~ ......

COCNl1W£ ArrRENTlCESlltP

Lepper and Greene (1979) discuss theimportance of creating learning environ~

ments in which students perform tasksbecause they are intrinsically related toa goal of interest to them, rather thanfor some extrinsic reason, like gettinga good grade or pleasing the teacher.In reading and writing, for example,intrinsic motivation might be achievedby having students communicate withstudents in another part of the world byelectronic mail.

4. Exploiting cooperation refers to having stu·dents work together in a way that fos·ters cooperative problem solving. Learn·ing through cooperative problem solvingis both a powerful motivator and a pow­erful mechanism for extending learningresources. In reading, activities to exploitcooperation might involve having stu·dents break up into pairs, where one stu·dent articulates his thinking process whilereading, and the other student questionsthe first student about why he made dif­ferent inferences.

Themes in Research on CognitiveApprenticeship

In the years since cognitive apprenticeshipwas first introduced, there has been exten­sive research toward developing learningenvironments that embody many of theseprinciples. Several of these principles havebeen developed further; in particular, situ·ated learning, communities ofpractice, com·munities of learners, scaffolding, articula­tion, and reflection.

Situated Learning

Goal·based scenarios (Schank et aI., 1994,Nowakowski, et a1., 1994) embody manyof the principles of cognitive apprentice.ship. They can be set either in computer·based environments or naturalistic environ­ments. Learners arc given real-world tasksand the scaffolding they need to carry outsuch tasks. For example, in one goal.basedscenario learners are asked to advise married

J................

53

couples as to whether their children arelikely to have sickle-cell anemia, a geneti­cally linked disease. In order to advise thecouples, learners must find out how differ·ent genetic combinations lead to the dis·ease and run tests to determine the par­ents' genetic makeup. There are scaffolds inthe system to support the learners, such asvarious recorded experts who offer advice.Other goal-based scenarios support learnersin a wide variety ofchallenging tasks, such asputting together a news broadcast, solvingan environmental problem, or developinga computer-reservation system. Goal·basedscenarios make it possible to embed cogni·tive skills and knowledge in the kinds orcon·texts where they are to be used. So peoplelearn not only the basic competencies theywill need, but also when and how to applythese competencies.

Video and computer technology hasenhanced the ability to create simulationenvironments where students are learningskills in context. A novel use of videotechnology is the Jasper series developedby the Cognition and Technology Group(1997) at Vanderbilt University to teachmiddle·school mathematics. In a series offifteen to twenty minute videos, students areput into various problem·solving contexts:for example, deciding on a business planfor a school fair or a rescue plan for awounded eagle. The problems are quitedifficult to solve and reflect the complexproblem solving and planning that occursin real life. Middle-school students workin groups for several days to solve eachproblem. Solving the problems results in amuch richer understanding of the underly­ing mathematical concwts than the tradi·tional school.mathematl·cs problems.

These kinds of situated·learning tasks aredifferent from most school tasks, becauseschool tasks are decontextualized. Imag~

ine learning tennis by being told the rulesand practicing the forehand, backhand, andserve without ever playing or seeing a tennismatch. If tennis were taught that way, itwould be hard to see the point of whatyou were learning. But in school, studentsare taught algebra and Shakespeare without

TilE CAMBRlDCE HANDBOOK OF TilE l.EARNINC SCIENCESH

being given any idea of how they mightbe useful in their lives. That is not how acoach would teach you to play tennis. Acoach might first show you how to grip andswing the racket, but very soon you wouldbe hitting the ball and playing games. A goodcoach would have you go back and forthbetween playing games and working on par­ticular skills - combining global and situ~

ated learning with focustitllocal knowledge.The essential idea in situated learning is totightly couple a focus on accomplishing taskswith a focus on the underlying competenciesneeded to carry out the tasks.

Communities ofPractice

Lave and Wenger (1991; Wenger, iI~y8) havewritten extensively about communities ofpractice and how learning takes place inthese contexts. They introduced the notionof legitimate peripheral participation to des~

cribe the way that apprentices participatein a community of practice. They describedfour cases ofapprenticeship and emphasizedhow an apprentice's identity derives frombecoming part of the community ofworkers,as they become more central members in thecommunity. They also noted that an appren­ticeship relationship can be unproductivefor learning, as in the case of the meatcutters they studied, where the apprenticesworked in a separate room and were isolatedfrom the working community. Productiveapprenticeship depends on opportunitiesfor apprentices to participate legitimatelyin the community practices that they arelearning.

The degree to which people playa cen­tral role and are respected by other membersof a community determines their sense ofidentity (Lave & Wenger, 1991). The centralroles are those that most directly contributeto the collective activities and knowledge ofthe community. The motivation to becomea more central participant in a communityof practice can provide a powerful incentivefor learning. frank Smith (1988) argues thatchildren will learn to read and write if thepeople they admire read and write. That is,they will want to join the "literacy club" and

will work hard to become members. Learn­ing to read is part of becoming the kind ofperson they want to become. Identity is cen­tral to deep learning.

Wenger (1998) argues that people partici­pate in a variety ofcommunities ofpractice­at home, at work, at school, and in hob­bies. In his view a community of practice isa group of people participating together tocarry out different activities, such as garagebands, ham-radio operators, recovering alco­holics, and research scientists. "For individ­uals, it means that learning is an issue ofengaging in and contributing to the practicesof their communities. For communities, itmeans that learning is an issue of refiningtheir practice and ensuring new generationsof members. for organizations, it means thatlearning is an issue of sustaining the inter­connected communities of practice throughwhich an organization knows what it knowsand thus becomes effective and valuable asan organization" (pp. 7-8).

Communities ofLearners

In recent years there has developed a "learn­ing communities" approach to educationthat builds on Lave and Wenger's (1991)notion of a community of practice. In alearning community the goal is to advancethe collective knowledge and in that way tosupport the growth of individual knowledge(Scardamalia & Bereiter, 1994, this volume).The defining quality of a learning commu­nity is that there is a culture of learning,in which everyone is involved in a col­lective effort of unJerstanding (Brown &Campione, 19(6).

There are four characteristics that a learn­ing community must have (Bielaczyc &Collins, 1999): (1) diversity of expertiseamong its members, who are valued fortheir contributions and given support todevelop; (2) a shared objective of contin­ually advancing the collective knowledgeand skills; (3) an emphasis on learning howto learn; and (4) mechanisms for sharingwhat is learned. It is not necessary that eachmember assimilate everything that the com­munity knows, but each should know who

I

w'ithto a(

tureiog,edgEthatkno'

BopelmoeTheint~

kona w00

ogy,spel

overesEwhlulallealgcofoctaslindth,ofonde<

,e,a ~

"bn

'p't"in~

intfo''u,exsu1111

o\'th

"Ieo

g'thtil

< ...........:~~; ,"-

COGNITIVE APPRENTICESHIP 55

.,

,,-l<e>r

.0

,-:ew,gohn­>0

•

within the community has relevant expertiseto address any problem. This marks a depar·ture from the traditional view of school­ing, with its emphasis on individual knowl­edge and performance, and the expectationthat students will acquire the same body ofknowledge at the same time.

Brown and Campione (1996) have devel­oped a model they call Fostering a Com­munity of Learners (FCl) for grades 1--8.The FCl approach promotes a diversity ofinterests and talents, in order to enrich theknowlcdge of the classroom community asa whole. The focus of FCl classrooms ison the subject areas of biology and ecol­ogy, with central topics such as endangeredspecies and food chains and webs. There is anoverall structure of students (1) carrying outresearch on the central topics in small groupswhere each student specializes in a partic­ular subtopic area, (2) sharing what theylearn with other students in their researchgroup and in other groups, and (3) prcparingfor and participating in some Mconsequentialtask" that requires students to combine theirindividual learning, so that all members inthe group come to a deeper understandingof the main topic and subtopics. Teachersorchestrate students' work, and support stu­dents when they need help.

In the FCL model there are usually threeresearch cycles per year. A cycle begins witha set of shared materials meant to builda common knowledge base. Students thenbreak into research groups that focus on aspecific research topic related to the cen­tral topic. For example, if the class is study­ing food chains, then the class may breakinto five or six research groups that eachfocus on a specific aspect of food chains,such as photosynthesis, consumers, encrgyexchange, and so on. Students research theirsubtopic as a group and individually, withindividuals "majoring" by follOWing theirOwn research agendas within the limits ofthe subtopic. Students also engage in reg­ular "cross·talk~ sessions, where the differ­ent groups explain their work to the othergroups, ask and answer questions, and refinetheir understanding. The research activi­ties include reciprocal teaching (Palincsar &

Brown, 1984), guided writing and compos·ing, consultation with subject matter expertsoutside the classroom, and cross-age tutor­ing. In the final part of the cycle, studentsfrom each of the subtopic groups cometogether to form a Mjigsaw" group (Aronson,1978) in order to share learning on thevarious subtopics and to work together onsome consequential task. In the jigsaw, allpieces of the puzzle come together to forma complete understanding. The consequen­tial tasks "bring the research cycle to an end,force students to share knowledge acrossgroups, and act as occasions for exhibitionand reflection" (Brown & Campione, 1996,p. l 0 3)'

A key idea in the learning·communitiesapproach is to advance the collective knowl­edge of the community, and in that wayto help individual students learn. This isdirectly opposed to the approaches found inmost schools, where learning is viewed as anindividual pursuit and the goal is to trans·mit the textbook's and teacher's knowledgeto students. The culture ofschools often dis·courages sharing of knowledge - by inhibit·ing students from talking, working on prob·lems or projects together, and sharing ordiscussing their ideas. Testing and gradingare administered individually. When takingtests, students are prevented from relyingon other resources, such as other students,books, or computers. The whole approachis aimed at ensuring that individual studentshave aU the knowledge in their heads thatis included in the curriculum. Thus, theleaming.community approach is a radicaldeparture from the theory of learning andknowledge underlywg schooling.

'.

Scaffolding

Computer-based, interactive learning envi­ronments can be designed to offer supportto learners in various guises, so that stu­dents can tackle complex, difficult tasks.Scaffolding is the support a system providesto learners as they carry out different activi­ties (Wood, Bruner, & Ross, u)76). This cantake the form of structured or highly con­strained tasks, help systems that give advice

"Hf. CAMRRIOCE IIA1'OIlBOO" Of 1111:: Lf.ARNING SClt;~CF.S56

when the learner does not know what to door is confused, guided tours on how to dothings, hints when needed, and so on. Onefann that scaffolding takes is that the SYS*tern can do many of the low-level chores,such as arithmetic calculations, while thelearner concentrates on the higher-level taskof deciding what to do. Another form isthat the system can provide an overall struc­ture that allows completion of a complextask, guiding students to individual compo­nents of the task, an4- showing them howeach component fits into the overall task.Scaffolding helps learners carry out tasksthat are beyond their capabilities. Quintanaet a1. (:W04) suggest twenty specific strate­gies for designing scaffolds to support sensemaking, inquiry, articulation, and reflectionin computer-based learning environments.In most situations, scaffolding naturally fadesas learners are able to accomplish tasks ontheir own.

In an analysis of computer·based learn­ing environments, Reiser (2004) points outthat most of the work on scaffolding hasfocused on Sfnlcturing the task for students,in order to make it easier for learners toaccomplish the task. But he emphasizes thatthere is another important role for scaffold­ing - problematizin.g the student's perfor­mance, or explicitly questioning the key con­tent and strategies used during the task, sothat students reflect more on their learning.Although this may make the task more dif­ficult, it can facilitate learning.

Bruner based his concept of scaffoldingon Vygotsky's (1978) notion of the zon.e ofproximal development, which described howadults can support learners to accomplishtasks that they cannot accomplish on theirown. Hence, the focus of research on scaf­folding (see for example Davis and Miyake,2004) has been on supporting individuals intheir learning. But Kolodncr et 31. (2003)point out that it is important to scaffoldgroups as well as individuals. So, for exam­ple, in their work teaching science, theyfirst provide students with focused collab·oration activities to solve simple problems,which they caU "launcher units." Engaging inthese activities and reflecting on them helps

students to collaborate more effectively andto understand the value of collaboration.

In schools, needing to ask for extra helpoften implies that the student is inferior,so students are reluctant to ask for help.When scaffolding is provided by computers,it comes without criticism and without oth·ers knowing that the student needed help.Computers offer a kind of scaffolding thatavoids stigmatization and provides individu­alized instructional support.

Articulation

In order to abstract learning from particularcontexts, it is important to articulate one'sthinking and knowledge, so that it becomesavailable in other contexts. There have beenseveral successful examples ofhow effectivegroup discussions can be in classrooms. Forexample, Lampert (Lampert, Rittenhouse,& Crumbaugh, 1996) showed how fifthgrade children can form a community ofinquiry about important mathematical con­cepts. She engaged students in discussionof their conjectures and interpretations ofeach other's reasoning. Techniques of thiskind have been successful with even youngerchildren (Cobb & Bauersfeld, \(}9S) andmay partly underlie the success of Japanesemathematical education (Stigler & Hiebert,'999)·

A notable method for fostering articula­tion in science is the Itakura method devel­oped in Japan (Hatano & Inagaki, 1991).First, students make different predictionsabout what will happen in a simple exper­iment, where they are likely to have differ­ent expectations. For example, one experi­ment involves lowering a day ball into waterand predicting what will happen. After stu­dents make their initial predictions, they dis­cuss and defend among themselves why theythink their predictions are correct. After anyrevisions in their predictions, the experimentis performed and discussion ensues as to whythe result came out the way it did.

Sandoval and Reiser (2004) have devel­oped a computer system called the Biol­ogy Cuided Inquiry Learning Environment(BGuILE) that supports students in making

st,\t

ccs1cb(,

Priptl

"eon,te(c

.•'. . .t~· ;;

COGNITIVE APPRl:::/'ITICF-SHIP

,d

Ip>C,

,p.rs,h­Ip.,atlu-

liarle'sneseentiveFo'use,fifthy ofcon­;stanIS ofthis

IOgerand

mese:bert,

lCula­level­1991).:tions~xper­

differ­xperi­water

er stu­ey dis­Iy theyter anyrimentto why

devel­e Biol­)Omentmaking

scientific arguments in the context of pop­ulation genetics. The system presents thestudents with a mystery of why many ofthe finches in the Galapagos Islands diedduring a period of drought (see Edelson &Reiser, this volume). In order to solve themystery, students have to analyze extensivedata that were collected by scientists andcome up with a reasoned conclusion as towhy some finches died while others sur­vived. The Explanation Constructor tool inthe system prompts the students to put inall the pieces ofa sound genetics-based argu­ment, after they have decided what causedthe finches to die. Hence, the system scaf­folds students to articulate their argument ina much morc explicit form than they wouldnormally do.

The Knowledge Forum environment dev­eloped by Scardamalia and Bereiter (thisvolume; 1994) is an environment where stu­dents articulate their ideas in writing overa computer network. The model involvesstudents investigating problems in differentsubject areas over a period of weeks ormonths. As students work, they enter theirideas and research findings as notes in anonline knowledge base. The software scaf­folds students in constructing their notesthrough features such as theory-buildingscaffolds (e.g., "My Theory," "I Need toUnderstand'1 or debate scaffolds (e.g., "Evi­dence For''). Students can read through theknowledge base, adding text, graphics, ques­tions, links to other notes, and commentson each other's work. When someone hascommented on another student's work, thesystem automatically notifies them about it.The central activity of the community iscontributing to the communal knowledgebase. Contributions can take the form of(a) individual notes, in which students stateproblems, advance initial theories, summa­rize what needs to be understood in order toprogress on a problem or to improve theirtheories, provide a draWing or diagram, andso on; (b) views, in which students or teach­ers create graphical organizations of relatednotes; (c) build-ons, which allow studentsto connect new notes to existing notes' and(d) "Rise Above It" notes, which synth~ize

57

notes in the knowledge base. Any of thesekinds of contributions can be jointlyauthored. The goat is to engage students inprogressive knowledge building, where theycontinually develop their understandingthrough problem identification, research,and community discourse. The emphasis ison progress toward collective goals ofunder­standing, rather than individual learningand performance

Reflection

Reflection encourages learners to look backon their performance in a situation, andcompare their performance to other perfor­mances, such as their own previous perfor­manct~s and those of experts. Reflection hasreceived much attention as a vital aspect ofthe learning process for both children andadults. Schon (1983) describes how system­atic reflection on practice is critical for manyprofessionals engaged in complex activities.Designers of learning environments oftenbuild supports for reflection into tasks byasking students to discuss and reflect uponthe strategies used to guide their actions.Reflection can highlight the critical aspectsof a performance and encourage learners tothink about what makes for a good perfor­mance and how they might improve in thefuture.

There are three forms that reflection(an take, all of which are enhanced bytechnology: (I) reflection on your process,(2) comparison of your performance to thatof others, and (';) comparison of your per­formance to a set of criteria for evaluatingperformances:

~

• Reflection O~l your process: Because tech·nology makes it possible to record perfor­mances, people can look back at how theydid a task. One useful form of reflectionis an ~abstracted replay," where the crit­ical decisions made are replayed. A sys­tem that teaches complex problem solv­ing could allow learners to compare theirdecisions in solving a complex problemto an expert solution, so that they can seehow they might have done better.

58 THF. CAAIBRIOCE IIANDROOK OF THE LEARNIl"G SCIE:"Ct,S

• Comparison ofyour perfonnance ro thac ofothers: One of the most effective waysthat people learn is by comparing differ­ent performances, including their own,to determine what factors lead to sue·cess. This is called "perceptual learn­ing~ (Bransford et aI., 1989). Technologymakes it possible to record different per­formances that learners ~an then analyze.

• Comparison ofyour perfonnance to a set ofcriteria for evaluaringperformances: One ofthe most effective ways to improve per­formance is to evaluate how you did withrespect to a set of criteria that determinegood performance. For example, Whiteand Frederiksen (1998) showed that stu­dents who evaluated their performanceon projects using a set of eight crite­ria improved much more than studentswho carried out the same tasks, but didnot reflect on their performance in thesame way. In fact this reflection helpedthe weaker students much more than thestronger students.

The essential way people get better atdoing things is by thinking about what theyare going to do beforehand, by trying to dowhat they have planned, and by reflectingback on how well what they did came oul.If they can articulate criteria for evaluatingwhat they did, this will help them as theyplan what to do on the next cycle. The wideavailability of computers and other record·ing technologies makes performances easierto produce and to reflect on. For example,students can now produce their own newsbroadcasts, musical performances, or plays,either on audiotape, videotape, or cable tele·vision, and send them to other schools orto parents. Furthermore, they can play theseback, reflect upon them, and edit them unlilthey are polished. One of the best examplesof the use oftechnology for recording perfor­mances has been in Arts Propel (Gardner,1991) with its cycle of performing, reflect­ing upon the performance in terms of a setof criteria, and then performing again. Mosteducational practice has not recognized thepower of this learning-cycle approach.

Conclusion

A5 these examples illustrate, there has beenextensive research over the last fifteen yearsthat has incorporated the principles of cog­nitive apprenticeship in the design of learn­ing environments. As computer-based learn­ing environments become more pervasive,there is likely to be continued developmentof new ways to embody these principles intheir design.

References

Aronson, E. (H)78). Tlrejigsawclassroom. BeverlyHills, CA: Sage.

Bielaczyc, K., & Collins, A. (1999) Learning com­munities in classrooms: A rcconceptualizationof educational practice. In C M. Rcigcluth(Ed.): Instructional-design theories and models: Anew paradigm 0/ instructiollal theory (pp. 269­ZQ2). Mahwah, NJ: lawrence Erlbaum Ass0­ciates.

Bransford, J. D., Franks, J J, Vye, N. J, &Sherwood, R. O. (1989). New approaches toinstruction: Because wisdom can't be told. InS. Vosniadou &A. Ortony (Eds.), Similarity andanalogical reasonillg (pp. 470-497). New York:Cambridge University Press.

Brown, A., & Campione, 1. (1(~96). Psychologicaltheory and the design of innovative learningenvironments: On procedures, principles, andsyste;:ms. In l. Schauble & R. Glaser (Eds.) 111110­

vatiOIlS in learning: New environments for edu­cation (pp. 289-325). Mahwah, NJ: LawrenceErlbaum Associates.

Brown, J S., Collins, A., & Duguid, P. (1989).Situated cognition and the culture of learning.Educational Researcher, 18(1), p-.p.

Burton, R., Brown, J. 5., & Fischer, G. (1984)' Ski­ing as a model of instruction. In B. Rogoff and JLave (Eds.), Everyday cognition: Itsdevdopmen­tal and social cvntext (pp. 1,9-15°). Cambridge,MA: Harvard University Press.

Chi, M. T. H., Bassok, M., Lewis, M. w.,Reimann, P., & Glaser, R. (1989). Self­Explanations: How students study and useexamples in learning to solve problems. Cog­lIitive Science, J3, 145-182.

Cobb, P., & Bauersfe1d, H. (Eds.) (1995). Theemngena of mathematical meaning: Inleraaiml

•

ill c.Eric

Cogni(1 "C;illstlmenl·iat.

Collin

'"M"for iYrorl

Collin.(1<.)8;:heics.illg,Gla.Erlb

C.,1Iin:Ae<studKna,,",IBass

Collin~

pre.\(Oct!rmdIh,)

Collin~

tht'QluthIIwil,

NJ:

Cuhant.onl

Da\·is,dalSrin

G.lrJn,

alterf(lrdChaillsrr

H:ltamthr01

1.. R.spr"Cl

HX)"::11 !'

Kolnclr1. T.,

COGNITIVE APPRENTICF.SHIV ;9

,y

,­'"thA,­,,-

&toIn

,wl,rk:

icalling

'ndmo­~du-

:nce

8q)_ling.

Ski­.oJJ.men­idgl'.

W.Self­

I uscCog-

I. ThelUtiutl

in classroom cultures. Mahwah, NJ: LawrenceErlbaum Associates.

Cognition and Technology Group at Vanderbilt(1997). The Jasper Project: Lessons in cumculum,ins/nlerion, assessment, and professional develop­ment. Mahwah, NJ: Lawrence Erlbaum Asso­ciates.

Collins, A., & Brown, 1. S. (1988). The computeras a tool for learning through refleaion. In H.Mandl and A. Lesgold (Eds.), Learning issuesfor intelligent tutoring systems (pp. l-t8). NewYork: Springer.

Collins. A.. Brown, 1. S., & Newman, S. E.(H)89). Cognitive apprenticeship: Teachingthe craft of reading, writing, and mathemat­ics. In L B. Resnick (Ed.), Knowing, learn­ing, and instruction: Essays in honor of RobertGlaser (pp. 4,';-494)' Hillsdale, NJ: LawrenceErlbaum Associates.

Collins, A., Hawkins, 1., & Carver, S. M. (1991).A cognitive apprenticeship for disadvantagedstudents. In B. Means, C. Chelemer & M. S.Knapp (Eds.) Teachillg advanced sl...'ifls to at-risksrudlmts. (pp. 216-243). San Francisco: Jossey­B",_

Collins, A., & Smith, E. E. (1982). Teaching theprocess of reading comprehension. In D. K.Detterman & R. 1. Sternberg (Eds.), How muchand how can intellig/!"tlce be increased? (pp. 173­18»). Norwood, NJ: Ablex.

Collins, A., & Stevens, A. L. (l\}83). A cognitivetheory of interal".""tive teaching. In C. M. Reige­luth (Ed.), Instructional thsign fheories andmodels: An 0VfmJiew (pp. 247-278). Hillsdale,NJ: Lawrence Erlbaum Associatt's.

Cuban, I.. (19S4). Huw teachers faught. New York:Longman.

Davis, E. A., & Miyake, N. (Eds.) (2004). Spe­cial issue: Scaffolding. JOUTTUJ.l of the Leami'lgSciences, q ()), 265-4) l.

Gardner, H. (1991) Assessment in context: Thealternative to standardized testing. In B. Gif­ford & C O'Connor (Eds.), FutureassessmeJlts:Changing views of aptitude, cu:hievnnent, andinstruction (pp. 77-120). Boston: Kluwer.

Hatano, G., & Inagaki, K. (I(J91) Sharing cognitionthrough collective comprehension activity. In:L. Resnick, J. Levin, & S. D. Teasley (Eds.), Per­SjJ«tives on socially shared cognirion (pp. l31­H8). Washington, DC: American Psychologi­cal Association.

Kolodner.1. L., Crismond, D., Fassc, B. B., Gray,1. T., Holbrook, J., Ryan, M., & Puntambekar,

S. (200)). Problem-based learning meets case­based reasoning in the middle-school scienceclassroom: Putting a Leaming-by-~ign cur­riculum into practice. Journal of the LeomingSciePlces, 11(4),495-548.

Lampert, M., Rittenhouse, P., & Crumbaugh, C.(1t)96). Agreeing to disagree: Developingsocia­ble mathematical discourse. In D. Olson & N.Torrance (Eds.), Handbook of education andhuman develcpment (pp. 731-764). Oxford:Bla,kwell's Press.

Lave, J (1988). The culture ofacquisition and thepractice of understanding (Report No. IRLSH­0007)' Palo Alto, CA: Institute for Research onLearning.

Lave, 1., & Wenger, E. (191)1). Situated leaming:Legitimate /'(Iriph/:ral participation. New York:Cambridge University Press.

Lepper, M. R., & Greene, D. (1979)' The hid·den costs of reward. Hillsdale, NJ: LawrenceEtlbaum Associates.

Norman, D. A. (1973). Memory, knowledge. andthe answering ofquestions. In R. L. Solso (Ed.),Contemporary issues ill cognitive psycholoK)!: J111!Loyola symposium (pp. 1)5-16;). Washington,DC: Winston.

Nowakowski, A., Campbdl, R., Monson, D.Montgomery, J, Moffl,tt, C, Acovelli, M.,Sch:lnk, R., & Collins, A. (1994). Goal-basedscenarios: A new approach to professional edu­cation. Educational Technology, 301(9), ~-;~.

Palincsar, A. S., & Brown, A. L. (11)84). Recip­rocal teaching of comprehension-fostering andmonitoring activities. Cognition and Instruction,'(2),117-175.

Quintana, C, Reiser, B. 1., Davis, E. A., Krajcik,J, Fretz, E., Duncan, R. G., Kyza, E., Edel­son, D., & Soloway, E. (2004). A scaffoldingdesign framework fo software to support sci­ence inquiry. Journal of the Le"millg SciC/ICes,

13(,),337-386 .Reiser, B. 1. (2oo4).;;caffoldingcomplex learning:

The mC'Chanis~'ofstructuring and probl.."lTla­tizing student work. Journal of the Leaming Sci·

~,13(3), 273-,04'Sandoval, W. A., & Reiser, B. J (~oo41.

Explanation-driven inquiry: Integrating con­ceptual and epistemic scaffolds for scientifi,inquiry. Sciem:e Education, 88, H;-37~.

Scardamalia, M., & Bereitcr, C. (1994)' Com­puter support for knowledge-building commu·nities. Joumal of the Leamillg ScieJlces, 3h),265-28 3.

THE CAMlIRlDGE HANDllooK Of THE L.EARNING SCIENCES

-'

-~;.

."

60

Scardamalia. M., Bereiter, c., & Steinbach, R.(1984)' Teachability of reflective processesin written composition. Cognitive Science, 8.

i73-H}o.

Schank, R. c., Fano, A., Bell, 6., &Jo;'a, M. (1994)The design of goal-based scenarios. Journal ofthe Learning Sciences, 3(4), 30 5-34 6.

Schoenfeld. A. H. (1985)' Mathematical problemsolving. Orlando, FL: Academic Press.

Schon, O. A. (1983)' The reflective practitioner:How professionals think in action. New York:

Basic Books.Smith, F. (J<)88). Joining the literacy club.

Portsmouth, NH: Heinemann.

Stigler, J, & Hiebert, 1. (1999)' The teaching gap:Best ideas from the world's teachers for improv­ing education in the classroom. New York: Free

Press.

Vygotsky, L. S. (1978). Mind in society: Thedevel·opmenf of higher mental processes. (M. Cole,Y. John-Steiner, S. Scribner, & E. Soubcrman,Eds.) Cambridge, MA: Harvard University

Press.Wenger, E. (1998). Communities of practice:

Learning, meaning, and identity. New York:Cambridge University Press.

White, B. Y. (1984). Designing computer gamesto help physics students understand Newton'slaws of motion. Cognition and InstnlCtion, 1(1),

69-108.White, B. Y., & Frederiksen,J. R. (1998). lnquiry,

modeling, and meucognition: Making scienceaccessible to aU students. Cognifion and Instnlc·

tion, 16(1), 3-nS .Wood, D., Bruner, 1., & Ross, G. (1976). The role

of tutoring in problem solving. Joumal ofChildPsychology and Psychiatry, 17,89-100.