Вы находитесь на странице: 1из 31

Participatory Simulations: Building Collaborative Understanding through Immersive Dynamic Modeling Author(s): Vanessa Colella Reviewed work(s): Source:

The Journal of the Learning Sciences, Vol. 9, No. 4 (2000), pp. 471-500 Published by: Taylor & Francis, Ltd. Stable URL: http://www.jstor.org/stable/1466765 . Accessed: 31/10/2012 20:11
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.

Taylor & Francis, Ltd. is collaborating with JSTOR to digitize, preserve and extend access to The Journal of the Learning Sciences.

http://www.jstor.org

THEJOURNAL OFTHELEARNING SCIENCES, 9(4),471-500 Erlbaum Inc. Copyright Associates, ? 2000,Lawrence

Simulations: Participatory Building


CollaborativeUnderstanding Through ImmersiveDynamic Modeling
VanessaColella
MITMedia Laboratory Cambridge,MA

Thisarticleexplores a new way to helppeopleunderstand dynamic syscomplex, tems.Participatory simulations intolife-sized, plungelearners computer-supported newpaths to scientific simulations, creating small,comunderstanding. By wearing called aretransformed intoplayers in municating computers Thinking Tags,students a large-scale microworld. Likeclassicmicroworlds, simulations create participatory a scenario, mediated thatenables andexperimenrules, by a setofunderlying inquiry tation. Inaddition, thesenewactivities allowstudents to "diveinto" a learning environment anddirectly withthecomplex athand. Thisarticle describes engage system andanalyzes a set of participatory simulations thatwereconducted witha group of school Thestudents' aretracked their initial from high biologystudents. experiences encounter withanimmersive simulation theirexploration of thesystemand through finaldescription of itsunderlying rules. Thearticle theeducational explores potential of participatory simulations. Theresults of thispilotstudysuggest anopportunity to further therolethatpersonal canplayin developing investigate experience inquiry skillsandscientific understanding. The studentsin a science classroomarechattering away as they play with the latest simulation. A is virus about to out a small community.Will the incomputer wipe habitantsdiscover a way to survive?A small groupof studentsin one cornerstare intentlyat a computer,waiting for the results.As they wait, the virusmysteriously infects a few "players" on the otherside of the classroom.Shrieksecho throughthe room as each new set of red lights indicatesthat anotherplayerhas succumbedto

andrequestsforreprintsshouldbe sentto VanessaColella, MITMediaLaboratory, Correspondence 20 Ames Street,E15-120H Cambridge,MA 02139. E-mail:vanessa@media.mit.edu

472

COLELLA

thedisease. Without Each disease. toevade thespreading warning, player struggles redlightsemblazon thewholepopulation. Thediseasehasrunits course. Thinkfora moment about theimagethatstoryconjures up foryou.If youpicturedthis gameunfolding, huddled you mighthavepictured groupsof students arounda desktopcomputerplaying the latest simulationgame-a sort of "SimVirus" ornewvirtual satcloseto the a fewstudents Outbreak. reality Perhaps fell ill. Perhaps a monitor whileothers themas their around behind players jumped few foughtfor controlof the mouseas they triedin vain to save theirplayer. Children andthendecide sucha gamewouldobserve theresults on screen playing howto use thatinformation model. the simulation to better understand Muchof ourimagination newkinds canbe usedto enable about howcomputers of learning in thesciencesis constrained motifof thecombytheboxandmonitor atleastnot thegamedescribed ona computer, aboveis notplayed However, puter. in which a traditional Thisarticleexplores simulations, computer. participatory students becomeplayers in unique, life-sizedgamesthataresupported by small, wearable computers. screenand simulations takethe simulation off of the computer Participatory Thestudents abovearenotjust worldof thelearner. bringit intothe experiential thesimulation; in a veryrealsense,theyare thesimulation. watching By wearing smallcomputers calledThinking eachbecomeagentsin the Tags,the students simulation. Thestudents do notneedto struggle to keeptrackof whichplayeris thatfolredlightsbelongto theirclassmates. Thequestions sick,forthe flashing low-Who gotthemsick?When? of examining How?Why?-are notmerely part a computer of their the underlying model;theyarepartof discovering mysteries veryownviralepidemic. in which simulations buildonthecharacteristics of microworlds, Participatory modelscanbe executed, andaugment of realworldexthemwiththeaffordances Thesenew environments area kindof role-playing perience. gamethatcombines the immediacy of real-lifeadventure of with the consistent rulesand structure microworlds. of a a simulation sysParticipants experience computer-supported temandthencollaboratively Inkeeping its dynamics. withthecallsforinexplore science, developingskills for systems thinking,and fostering quiry-based collaborative in scienceclasses(National on ScienceEducaCommittee learning tion Standards andAssessment, this 2061, 1996;Project explores 1993), project howlearning simulation. takesplaceintheenvironment created by a participatory DESIGNING EXPERIENCES
There is a long history of theoretical claims that children construct their own knowledge through experience (Dewey, 1916, 1988; Montessori, 1912; Papert, havetakenup the taskof designingeducative 1980;Tanner,1997). Many educators

UNDERSTANDINGTHROUGHDYNAMIC MODELING

473

experiences, often selecting or creatingparticularmaterialsto enable an experience. When developing his concept of kindergarten, FriedrichFroebel pioneered the idea thatparticular which he called objects, "gifts,"couldbe given to childrento stimulatecertainkindsof exploration.He arguedthatthese gifts would provideexperiences for childrenthatwould likely lead to certainkinds of cognitive development (Brosterman,1997).1Much of his notion of kindergarten focused on how the build of the children to would enable knowledgein a coherent orderlydelivery gifts fashion. Years later, Vygotsky wrote extensively on the notion that tools (like Froebel'sgifts) could enrichandbroadenboththe scope of activityandthe scope of thinking of the child (Vygotsky, 1978). Otherresearchershave even speculated aboutthe ways in which the objects presentin the environmentcould actuallyinduce development(Fischer, 1980).2 Computersfit right into this lineage. Even before the prevalence of personal computers, Seymour Papertenvisioned a future in which computer-basedtools would providechildrenwith a whole rangeof transformative developmentalexpewithinthese powerfulcomriences(Papert,1980). He imaginedthatconstructions puting engines would become fodder for children's imaginativeand intellectual ruminations,much like gears (his own childhoodtool) had become for him. The fact that computerscould take on so many differentroles, potentially a role per child, was especially exciting. Much effort has been expended to build computationaltools that provide opportunitiesfor childrento engage in experiences that would not be accessible to children without those tools (Resnick et al., 1998). Virtual communities offer places for children to construct alternate realities (Bruckman, 1998); computer-basedmodeling environmentsenable the design and constructionof comlabs plex papersculptures(Eisenberg& Eisenberg, 1998); microcomputer-based facilitate children's collection of scientific data (Tinker, 1996); and Newtonian-basedenvironmentsallow explorationof the laws of physics (White, 1993). Each of these computerized tools supports exploration, investigation, or creation-activities central to an educative experience. The next section describes tools thatprovidedthe conceptualandcomputamicroworlds,the computer-based tional frameworksfor the developmentof a new class of educationalexperiences called participatory simulations.

A Computational "Sandbox"
Microworldswere originally conceived to give childrena sort of computational sandbox-a world in which they could manipulateobjectson the computerscreen.
'See also Lillard(1972) for relatedwork. see Nor2For anotherperspectiveon the importanceof tools in the developmentof understanding, man (1993).

474

COLELLA

In a real sandbox,childrenuse buckets,shovels, and sandto createminiaturecastles. While creating these sandcastles,childrenoften grapplewith concepts like shape and scale. What base supportsthe tallest sandcastle?How big should two pebbles be if they aremeantto representa prince and a princess?A computerized sandbox offers more thanjust a sandboxon a screen. In a microworld-as in the real world-a child can take actionsthathave discernibleeffects on the world. But in a microworld,the child also has some access to the formalrulesthatgovernhis or heractions.Microworldsoffer a nonformalentryinto a worldbasedon formal,logical constructs. Picture a girl playing with a toy horse in her room. She can move the horse aroundand even have it "talk"to other animalsin the barnyard. The horse might as she altersthe speedwith which she flies the horse aroundher "gallop"and "trot" play space. In a microworld,her horse could still move aroundin space, talkingto otheranimals,but she mightbegin to investigatethe mathematical relationshipbetween the horse's two speeds. Dependingon the microworld,the computermight even show her an equationthatrelatesthose speeds. Or she could makethe galloping speed dependenton the trottingspeed. Certainly,she could perform similar mental operationsin the real world, but the microworldcan provide a seamless transitionfrom the nonformal,naive operationsin the real world to the formaldescriptions and investigations of those operationsin the microworld.In fact, research has suggested that microworldswhose formal descriptionsclosely mirror children's experience with patternsand activities can be betterlearningenvironments (diSessa, 1988). Most often, a microworldfocuses on a specific set of formalrules, constraining the types of actions a child can take but providing an opportunityto learn more about the rules governing those actions. Roschelle (1996) describes one such of the Envilearning activity, duringwhich two girls build up an understanding a microworld that facilitates of sioning Machine, exploration velocity andacceleration. Like many microworlds, the Envisioning Machine provides "an intermediate level of abstraction fromthe literalfeaturesof the physical world"(p. The becomes a 241). computer bridge linking the patternsand activities in the microworld(in this case, motion of a ball or particle)with the formalexpressionof those patternsand activities (arrows representingvelocity and acceleration)by connecting patternand activity to representationsof the underlying processes. This bridge enables childrento interactwith both the processes and patternsthey observe andthe formalsystems thatgovernthose patternsandprocesses.As much as Froebel's gifts facilitatedspecific activitiesand,in so doing,helped childrendevelop new understandings,microworldscan broadenthe range of activities and thoughtsin which childrencan engage.

Benefits of microworlds. Teaching andorganizing ofteninvolvescreating


special experiences to help children learn certain ideas. The flexibility of

UNDERSTANDINGTHROUGHDYNAMIC MODELING

475

microworldenvironmentsopens up the range of possible experiencesthat can be created.Some researchers have claimedthat"thecomputeris ... more flexible and in precise craftingexperiencesthatcan lead to essentialinsights"(diSessa, 1986, p. have constructedmicroworldsthat make possible 224). Teachersand researchers countless experiences,from exploringgeometricrelationshipsto buildinginteractive riverecosystems. Forexample,differentmicroworldsenable childrento focus an explorationon particular aspectsof physics (The EnvisioningMachine),mathematics (Logo), or politics (SimCity).One class of microworldsthatenablefocused explorationof complex, dynamicsystems,has gainedmainstream popularityin the few Game software like and SimLife past years. (1992) helped SimCity (1993) generate popularinterestin complex systems. Programslike Model-It (Jackson, Stratford, Krajcik,& Soloway, 1994), Stella (Roberts,Anderson,Deal, Garet,& Shaffer, 1983), StarLogo (Resnick, 1994), and Sugarscape (Epstein & Axtell, 1996) enable users to experimentwith complex systems and develop better intuitions aboutthe mechanismsthat govern dynamicinteractions. Microworlds let children experimentwith real concepts in play space, or as Pufall (1988) said, they create "a context within which childrencan think about discretespace as real and not aboutdiscretespace as an abstractionfromthe analogue worlds of sensory-motorexperience"(p. 29). With microworlds,learning experiences are no longer constrainedby what the real world has to offer. We can both limit and augmentthe real world, sometimes creatingsimplified spaces for exploringcomplex topics, othertimes creatingwholly new experienceson-screen. Pufall (1988) furtherspeculated that the new interactionsmicroworlds enable might "alter[children's]patternsof developmentby allowing [them]to interactin ways [they] cannot interactwith the 'real' world"(p. 21).

forlearnintroduced benefits Buildingon microworlds. Microworlds many


ing and presented some new challenges as well. Withouttrying to exhaustively coverthe benefitsof learningin the physicalworld,it is worthmentioningthatthere are humanties to interactionsin real space that are lost in cyberleamrning. Though some usersbecome enamoredof the machine(Turkle, 1984), othersfeel distanced fromthe patternsandprocessesthey observe on a computerscreen.For some people, this distance leads to a general distaste for the "cold,"unemotionalworld of computing(Turkle& Papert,1992). Othersareinclinedto believe everythingthey see on a computer,not questioningthe validityor appropriateness of simulationresults. Sociologist Paul Starr(1994) witnessedone user's lack of intellectualcuriosity about the underpinningsof SimCity and anothergroup's disinterestin rigorously questioning the assumptions underlying a computer model designed to forecast futurehealth care costs. In SimCity, the underpinningsof the model are hidden from the user, perhapsstifling curiosity.But the assumptionsin the health caremodel were readilyaccessible, suggestingthatdevelopinga full understanding of a computermodel is a formidabletask.

476

COLELLA

As much researchon microworldshas shown, these challenges are not insurmountable. Many microworld environmentsengage students in deep reasoning and sophisticatedanalysis (e.g., Eylon, Ronen, & Ganiel, 1996; Goldman, 1996; Papert, 1980; Roschelle & Teasley, 1995; Rothberg & Awerbuch, 1994; Schoenfeld, 1990; Tabak& Reiser, 1997; White, 1993). Microworldsenable a diverse set of experiences, encouragingchildrento broadenthe scope of their intellectual investigations.Effective microworldsdon't turn learners' "experience[s] into abstractions. like the laws of physics, into ex[Instead,they turn]abstractions, perience"(diSessa, 1986, p. 212). By actualizingthese experiences,microworlds enable learnersto directlyexperiencesimulations.Or more precisely, they enable users to enjoy experienceswith those simulationsthatareas directas we can make them (diSessa, 1986). In the past, directinteraction meantmanipulating with a simulatedenvironment agents or parametersin a microworldor controllingan avatarin a virtualworld. New technology allows us recast the notion of "directly"interacting with a computationallysimulatedexperience.We can now deploy simulationsin the real world, facilitatinga more personalexperiencefor learners.Ouraim is thatjust as microworldshave greatlyenhancedthe learningexperiencesavailableto students, simulationswill provideanotherrangeof learningexperiences,upon participatory which studentsand teacherscan draw.

faciliAnotherway to learn fromexperience. Participatory simulations


tateanotherway for learnersto collaboratively between investigatethe relationship and in rise to those the world and the rules that patterns processes give patternsand in of simulations characteristics build on the microworlds, processes. Participatory which models can be executed, and augment them with the affordancesof real world experience, enabling learnersto become the participantsin computer-supportedsimulationsof dynamicsystems in real space. Small, distributed computers createa life-sized microworldby deployingconsistent,computational rules in real space. Learnerscan experienceandinfluencethis simulationdirectly.This interaction, thoughstill mediatedby technology,is qualitativelydifferentfromothertechnology-controlledrole-playinggames that facilitateinteractionthroughavatarsor with the componentsof a microworld.Participants' personalconnectionsto the educationalsituationenable them to bring theirprevious experiencesto bear during the activity, establishstrongconnectionsto the activity and the otherparticipants, and, we hope, drawupon their experiencein the future.

Activities Participatory
The Participatory SimulationsProjectinvestigateshow direct,personalparticipation in a simulationleads to a rich learningexperiencethatenables studentsto explore the underlyingstructureof the simulation.The idea to use direct, personal

UNDERSTANDINGTHROUGHDYNAMIC MODELING

477

to help children(or learners)gain a new perspectiveor build a better participation is not a new one. Dewey emphasizedthe value of personalparticipaunderstanding tion in educativeexperiencesthroughout Inthe social sciences, perthe curriculum. spective-takingactivities are quite common (Seidner, 1975). Studentsmight be askedto takeon the role of communityactivistsor politiciansandsimulatea debate on the futureof the logging industry.This debategives the participants a way to representthe characters andthinkabouthow the variouscharacters mightfeel aboutan issue. Activities like these are less common in the sciences, where the mechanismsto be studiedare not humanfeelings andbehaviorbut concepts like planetarymotion or molecularinteractions. Nonetheless, studentssometimes take on those kinds of roles as well, perhapspretendingto be planets in orbit, in an effort to illustrate those phenomena.However, these activities are very different from their social science counterparts. Althoughthe social science activitiesmighthelp the students to think abouthow a politician, for instance,would feel and behave undercertain circumstances,the science activitiesdon't necessarilyhelp studentsto thinkabout the underlyingmechanismsof processes like planetarymotion. Role-playing activities attemptto create links between personal experience and a deeper understandingof why that experience happened,yet the science-based activities often end up being little more than large-scaleillustrations. Researchershave attemptedto connect personal and physical interactionsto underlying(nonhuman)mechanisms in a variety of ways. Papert(1980) tried to forge links between humanactionandthe rules of TurtleGeometryby askingchildren to pretendthey were the turtle and then translatethat understandinginto a of the instructionsfor the turtle'smovement.Resnick and symbolic representation Wilensky (1998) expandedon this idea, involving largegroupsof people in activities to help them gain a richerunderstanding of the rules governingemergentsystems. Recently,Wilensky and Stroup(1999) developeda networkarchitecture that students control over individual in a simulation environment. Regives agents searchersin systems dynamics also use group activities to help learnersdevelop systems thinking capabilities (Booth Sweeney & Meadows, 1995, 1996; Meadows, 1986; Senge, Roberts, Ross, Smith, & Kleiner, 1994). Participatory simulationsbuild on microworldsandthese groupactivities,using wearablecomputersto createan explicit link between personalexperiencein real space and the underlyingrules thatmediatethose experiences(Colella, 1998; Colella, Borovoy, & Resnick, 1998).

THE PARTICIPATORY SIMULATIONS PROJECT


The Participatory SimulationsProjectlooks specificallyathow a new kindof learnenvironment can motivate learners,facilitate data analysis, collaborativetheing and of scienory-building experimentaldesign, and lead to a richerunderstanding

478

COLELLA

tific phenomenaand the processes of scientific investigation.By involving a large numberof studentsin a physical experience of a simulation,the projectbrings a microworldoff the computerscreenand into the participants' space. Ouraim is to establishwhether,and if so how, participation in these activities promotesthe development of both the motivationand ability to engage in scientific thinking. The Participatory SimulationsProjectis an extendedresearchendeavor,studysimulations can ing how personal explorationof life-sized, computer-supported Thousandsof help participants develop inquiryskills andscientific understanding. in variousactivitiesat schools, in workshops,andat conpeople have participated ferences. In each activity,people use small, wearablecomputersto become agents in the simulation.For instance, duringthe pond ecology simulationsome participantsbecome schools of"big fish," and othersbecome schools of"little fish." As they interactwith one another,the big fish "eat"little fish, and the little fish "eat" fish food. The Tags keep trackof the numberof fish in each school. Participants collaborativelyinvestigatethe ways in which cooperativeand competitivebehaviors alterthe dynamics of the ecological system and change the carryingcapacity of the pond. Similarly,the tit-for-tatgame allows participants to experiencegame from a can theory first-personperspective.Together,they explore how cooperative behaviorsevolve over time. In the virus activity describedin this article,participantsinteractas a disease moves throughtheircommunity.The groupworksto analyze the disease dynamicsand establishhow the behaviorof individualsinfluences the outcome of the simulation.In all of these participatory simulations,people collaborativelyexplorethe system by changingtheirown behavior,collecting data,runningexperiments,and observingthe effect that theirbehaviorhas on the dynamics of the system.

Technological Support
We use small, wearablecomputerscalled ThinkingTags to enabledirectparticipation in the simulation.The Tags collect informationfor the participants (like how other have and them the state of other to many playersthey met) help playinterpret ers (e.g., whethersomeone is "sick"or "healthy"). Unlike the traditional notion of wearablecomputing,which focuses on connectingusersto an externalnetworklike the Web, the Tags connect all of the participantsin their own discrete network, which facilitatesinteruserconnectivityandprovidesthe computational supportfor the simulation.Ratherthanjust transforming the experienceof an individual,participatory simulationstransformthe interactionsamong people by linking them become througha personalizednetworkof communicating computers.Participants in a mediated of and their small, players computationally system comprised people personalcomputers. simulationsare supported Participatory by a variationof the ThinkingTag techat the Media Lab nology developed (Borovoy, McDonald, Martin,& Resnick,

UNDERSTANDINGTHROUGHDYNAMIC MODELING

479

each participant into an "agent"in a simula1996). The Tags areused to transform tion of a dynamic system. In these decentralizedsimulations,no one Tag acts as a server,andno large (traditional) computeris necessaryto run,experimentwith, or the We analyze system. developed a new version of the ThinkingTags3to facilitate collaborativeanalysis of many iterationsof the simulation.As in the original ThinkingTag design, we took careto ensurethatthe enhancedinformationdisplay would not interfere with participants' social interactions (Borovoy, Martin, Resnick, & Silverman, 1998; Borovoy et al., 1996; Ishii, Kobayashi, & Arita, 1994; Ishii & Ullmer, 1997). Like the originalThinkingTags, the Tags built for participatory simulationsare complete, albeit miniature,computerswith input and outputdevices and displays for the user. Each Tag possesses an infrared transmitter andreceiver,allowing it to dynamically exchange informationwith all other Tags in the simulation.As the simulationis running,the Tags are constantlyexchanging informationvia infraThe Tags have two disred, though this exchange is invisible to the participants. play devices, a double-digit numberpad and five bicolor LEDs (see Figure 1). Duringthe simulation,the information displayedon the Tags changes,andparticipants watch the Tags to discover informationabout themselves and about other players. A resistive sensor port acts as an input device, allowing users to attach small tools to theirTags and enablingthem to "dial-in"informationor change the programtheir Tag is running.This carefullychosen set of inputsand outputsprovides a rich set of user interactions, both duringthe simulationand duringthe subsequent analysis. Participants The 3-week-long pilot ParticipatorySimulations Study described in this article took place in anurbanpublichigh school classroom.All of the studentsvolunteered for the projectand were told that they would be participating in a projectto learn aboutdynamicsystems in science. Class time for 5 days over a 3-week periodwas devoted to activities associatedwith the Participatory SimulationsStudy. The chosen Biology class consisted mainly of 10th-gradestudentswho were describedby theirteacheras traditionally in science class. Sixteen poorperformers students-7 girls and 9 boys-participated in the study. The teacher also participated in the activities, and on Day 4, a studentteacherobservedthe class and participatedin the activities.The researcher (author)was the facilitatorof the classes. In addition,two studentsvideotapedthe activities.4

3Specialthanksto Kwin Kramerfor designing and building this version of the ThinkingTags. 4One studentwas a memberof the Biology class who preferred to not be filmed for religiousreasons and the otherwas a classmate from a differentbiology class.

480

COLELLA

1 TwovirusTags.ThetopTaghasmettwopeople FIGURE andis notsick.Thebottom Tag hasmetsix peopleandis sick,as indicated by thefive redLEDs.

Activities
Aside from a very brief introduction to the researcherand the basic operationsof the Tags, the students'first experiencein the Participatory SimulationsStudywas playing a disease simulationgame. The context was set for the first simulationby giving the studentsa challenge:to meet as manypeople as possible withoutgetting sick. They were told thatone of the Tags containeda virus andthatthey could elect to stop meeting people anytimethey wantedsimply by turningtheirTag aroundto face theirstomachs(orturningit off) andsittingdown. The studentswere told nothing abouthow the virus moved from one Tag to another,nor were they told anything aboutthe degree of contagiousness,the possibility for latency, or any other underlyingrule that could affect the spreadof the disease, leaving them in an ambiguous situation.None of the students'questions aboutthe behaviorof the virus were answered.Instead,they were given the opportunity to experienceandexplore the disease simulationfor themselves. The studentsparticipated for 45 to 55 min on each of 4 days and for 90 min on the last day. The projecthadthreephases. On the firstday (Phase 1), studentswere

UNDERSTANDINGTHROUGHDYNAMIC MODELING

481

introduced to the researcher andto a few otherexamplesof technologythatoperate on the same generalprinciplesas the Tags (Resnicket al., 1998). On Days 2, 3, and 4 (Phase2), studentsparticipated in disease simulations,or games, and in analyses of those simulations.This phase had three distinctcomponents:the initial disease simulation,the discussion of that simulation,and the developmentand execution of experimentsto test hypotheses about that simulation.The studentscompleted six disease games over the courseof the 3 days, with the discoveries from one simulationleading to the design of the next. Finally, on Day 5 (Phase 3), studentsreflected on their experiences in the Participatory Simulations Study and asked to participatein one final simulationgame.

The Disease Simulation


this pilot study,the underlyingrules of the simulation5 were kept conThroughout stant.The rules of the disease simulationwere as follows: * The virus was latent (invisible) for approximately3 min. * Any personwhose Tag hadthe virus, even if it was not visible, could infect anotherperson's Tag. * The probabilityfor infection when meeting an infected Tag was 100%. * People with Tags numbered1 or 2 in the ones position (1, 2, 11, 12, 21, etc.) were immuneto the virus. * ImmuneTags were not carriersof the disease. During the simulation,the numberpad displayed the numberof differentpeople with whom each participant hadinteracted, andthe five LEDs flashedredwhen the was sick. The also trackedinformationthatthe participants could access Tag Tags afterthe simulation,includingthe ID numbersof all of the individualswith whom a personinteracted,the time of all interactions,and the ID numberof the individual responsible for infection. During the study reportedin this article, students accessed the storeddataonly to confirmtheir final hypotheses. In some of our other simulationsprojects,this datahasbeen aggregated,displayedthrough participatory and used for more in-depthanalyses of disease transmission. StarLogo,

DataCollection
Bringing new computationaltools into a classroom can fundamentallyalter the structure of the class's interactions. The unit of analysis in the Participatory Simu5Sincethe Tags arefully programmable, these rulescan be modifiedor completelychangedfor a differentparticipatory simulation.Forexample,some disease simulationsconsist of a virus andanotheropportunisticinfection. Othersimulationsdiffermore substantially,like ourpond ecology game in which model predator-preyinteractions. participants

482

COLELLA

lations Studywas not the individualchild nor the individualchild plus the tool, but the whole cognitive system in the classroom (Newman, 1990; Salomon, 1993). Newman defines the cognitive system as follows: Theteacher a socialsystem creates intheclassroom certain kinds of disthatsupports courseandactivities; students collaborate withinthe system,contributing observaandconcrete suchastexts,projects, Thecognitive anddata. tions,answers, products theexternalized all of whichis proanddiscourse, systemincludes tools,texts,data, ducedby andfortheactivities. (p. 187) SimulationsStudy,attentionwas paidto how all aspectsof Duringthe Participatory the learningenvironment(the groupof students,theirconversations,and the tools they employed) contributeto building scientific understanding. This studyanalyzedconversationsandexplicit collaborativediscussionsduring the activities.The main sourceof dataforthe Participatory SimulationsStudywas a aimedto captureall of the completevideotapelog of the sessions that,in particular, whole-groupconversations.Inaddition,audiotape backupsweremadeof every sesthe project.Studentswere occasionsion, andfacilitatorlogs were keptthroughout ally askedto write down theirideas aboutthe simulationdynamics,andall of those studentresponseswerekept.We examinedthe datato findevidenceof ourfourmain aims:Duringthe activities,were studentsengagedinthe simulation? Couldstudents and evidence from Were able to the simulation? identify analyze designexperithey ments, predictoutcomes, and runexperimentsto confirmor deny theirideas? Did studentscarryout theirinvestigationsin a scientific manner?

ANALYSIS OF CLASSROOM ACTIVITIES Engagement in the Simulation


In the participatory simulationsProject,we aimed to motivate studentsby giving them a real experiencethatwas mediatedby a set of underlyingformalrules. One measureof success of a participatory simulation,then, is the extent to which students felt as though they actually experienced the simulation. In this case, we judged the experientialqualityof the simulationby observingthe extent to which studentssuspendedtheirdisbelief and actedas thoughthey were in the midst of an epidemic strikingthe membersof their small community. The following episode depicts some of the excitement and tension thatpermeated the learningenvironment: Episode 1 Researcher: I got it from her. Student: You all got the virus!

UNDERSTANDINGTHROUGHDYNAMIC MODELING

483

I'm dead. Stacy: Doug: (to Tony) Oh, you got the virus now. Tony: (looking at Tag) You got it started. Rick: (singing) I ain't got the virus. Student: I'm healthy. Meredith: (holding Tag up) I don't have the virus. Researcher: Who in this room met the most people? Chorus of students:I have 14, I got 16, I got 13 with no virus,me too, I got 14 with no virus. Student: I need some medicine. Studentsdisplayed a robust and persistentwillingness to suspend their disbelief and behave as though the simulationactivity was real. The learningenvironment promoteda strongconnectionbetweenthe studentsandthe simulation.WhenStacy exclaimed thatshe was "dead,"she was not talkingaboutan externalagent or avatar-she was talking about herself in the simulation.Similarreferencesoccurred throughoutthe study, as when a studentdeclaredthathe needed medicine. This level of engagement permeatedthe next 4 days of the project. As each game unfolded, the studentsonce again had a real-life experienceof an epidemic invading their small community.Theirtask was not to mentally constructthe dynamics of an epidemic from a written descriptionor a set of equations. Instead, they needed to figure out what was happeningin their community.The activity "arousedcuriosity,strengthened initiative,and set up desires andpurposes"in the of the simulationenvironstudents,propellingthem to develop an understanding ment (Dewey, 1988, p. 20). This compelling,interpersonal experienceis one of the simulationand set the stage for the learning key componentsof the participatory activities that followed. Though engagementin the immersive experience is an integraland important simulations,the immersivecomponentper se does not componentof participatory determinethe activity's educationalvalue. The experience's potentialfor leading to growthrests on its ability to allow the studentsto problematizetheirindeterminate situation(and later to inquireinto its underlyingstructure). In this case considerable learning occurred as students were able to step back from their immediateexperienceand analyzethe situation.Ackermann(1996) describedthis as studentsmove back and forth beprocess as "diving-in"and "stepping-out" tween full immersion in a problem and thinking about a problem. Similarly, Sterman(1994) distinguishedbetween the featuresof learning in and about dynamic systems.6Many scientific problemsoffer the chance to step outside of the for studyat a problemandthinkclearlyaboutit. Few problemsthatareappropriate
6See also diSessa (1986).

484

COLELLA

high school level offer the chance to dive so convincingly into a problem.Particifor studentsto enjoy both of these patorysimulationscreatea unique opportunity importantperspectivesduringthe processes of defining and solving problems. it has not alThe notion of diving into a scientific problemto betterunderstand ways been highly valued by researchers.The scientific communityhas traditionat the expense of more ally valued detached,objective modes of experimentation "connected"methods;however, some examples from scientific practiceindicate simulations thata revaluationof connectedscience may be in order.7 participatory can bring connectedscience to the classroomwithout forcing studentsto abandon the exploration of scientifically importantproblems. As students collected data and designed experiments,they remainedin touch with the problem at hand. A nontrivial characteristicof the participatory simulations environmentmade this connection possible-the studentswere collecting data about and experimenting on their own behavior.

Identification andAnalysis of Evidence


that Like many microworlddesigners,we wantedto createa learningenvironment enabledstudentsto define problemsand constructtestablehypotheses.In orderto begin the process of formally analyzing their simulation,the studentsneeded to a tractable problemandbegin identifyandanalyzeevidence to help themarticulate to hypothesizeaboutits structure. In this section, we describea few representative instancesin which they extracteddatafromthe simulationsand analyzedthatdata to betterframethe problemat hand. At the close of the first simulation,therewas no clearlydefinedproblemfor the studentsto explore, but they were certainlyin a problematicsituation.Almost all outcomefor many of the particof the studentsin the class were sick-a surprising The facilitatorasked if there who virus. that had avoided the thought they ipants was anyone in the class who managednot to get sick. The studentsbegan comparing notes in an attemptto explain the outcome of the simulation. First the students accumulateddata, and then they began to make assertions based on the available information.Some of the students' initial assertionswere hypotheses about why something happened, some were suggestions about how hypothesis, and otherswere ideas about they could prove or disprove a particular whatproblemthey shouldbe investigatingin the firstplace. Studentsoffered supevidence against many of these assertions. porting evidence for or contradictory As the available evidence accumulatedand ideas proliferated,the potential for constructingtestable hypotheses aboutthe viral behaviorgrew.
7SeeKeller(1983) for an example of how diving in to a problemcan yield innovativeandpreviously unimaginedsolutionsto scientific questionsandWilensky (1993, in press) for discussions of connected mathematicsand science.

UNDERSTANDINGTHROUGHDYNAMIC MODELING

485

In the following episode, studentswere presentingtheirpieces of datafromthe simulation.Notice that data in a participatory simulationare really observations about a student'sbehavioror state duringor afterthe game. Episode 2 Rick: Joan: We should all meet each other. I met Doug like 2 minutes before he gave the virus to other people and I didn't get sick. Allison: How do you clear these? I need a medicine, I need an antibiotic. Student: Researcher: Is thereanyonewho startedwith the virus otherthanthis guy in the front? Rick: Doug. (Supplyingthe name of the guy in front) Allison: That'sjust 'cause Doug's dirty. Joan: Doug didn't startoff with the virus. Researcher: Who startedout with the virus? Allison: 'Cause I met him, I met him. Joan: 'Cause I met Doug and I didn't get the virus. Allison: Doug was the second person I met. I ... I met her and then, I just, the virus was just like pop. Doug: Allison: I didn't get the virus until I got it from somebody else.

Here datawas presented(some of it before this episode began) that culminatedin the notionthatDoug hadinfecteda lot of people.But the students'suggestionswere not especially focused on runningexperimentsor testinghypotheses.Whenthe researcher restatedthe question"Whostartedoutwith thevirus?"the studentscontinued offering suggestions and ideas but did not responddirectlyto the question. A laterepisoderevealedthe students'morestructured to testthevalidity attempts he ofthe propositionthat last the couldbe person orshemet: infectedby aperson Episode 3 Liz: Liz: Rick: Stacy: Stacy: Allison: All right,I'm all set; I'm not meeting nobody else. I'm sick. Oh, I just boot beeped8her. Liz's the first one. Liz's the first one to get sick! Who'd you sharewith?9Do you remember? (While writing on the board)Wait, who was the last one you sharedwith?

8Becausethe Tags makea tiny "beep"each time they interactwith anotherTag, some studentsbegan describinga meeting as "beeping"or "bootbeeping."This languagewas laced with innuendoaboutthe type of interactionthat studentsfelt the Tags were simulating. with" is anotherway that studentstalk aboutmeeting one another. 9"Sharing

486

COLELLA

Liz: Allison: Stacy: Liz: Stacy: Liz:

Rick. Wait, you gotta go in order. OK, look at,Doug, Rickwas the lastpersonshe sharedwith. It's Rick's fault, it's all Rick's fault. No 'cause I sharedwith Rick. I sharedwith Rick too.

As students described their observations, like, "Rick was the last person she shared with," others responded either with data from their own experience or with hypotheses that could provide an interpretiveframe for the previous data. For instance, Liz hypothesized that "it's all Rick's fault" after a numberof observations that sick people had recently shared with Rick. This interpretive frame turned out to be inadequateto explain everyone's experience. Two students quickly rebuttedLiz's hypothesiswith observationsthat they had each met Rick and were not yet sick. At this point, studentsconvergedon a few problematicissues in their situation that they wanted to solve, including discoveringthe identity of PatientZero (the personwho startedout with the virus) anddescribingthe way thatthe virusmoved from one person to another.Their earlier,ill-structured presentationof evidence of evidence thatcould suggest to a more collection fragmentsgave way systematic which hypotheseswarrantfurtherinvestigation.The participatory simulationprovided a settingfor the studentsto engage in inquiry.Theirpatternof inquiryis consistent with the notion thatideas lead to more directedobservation,which in turn brings new facts to light and suggests fruitful directions to pursue (Dewey, 1938/1998).

Experimental Designand Execution


In addition to enabling students to identify problems and constructhypotheses, we designed the participatory simulationsenvironmentto facilitateexperimental and execution. Students design explore the underlyingrules of the simulationby their own behaviors and altering observing the effects of those alterationson the and outcome of the simulation. Like scientists probinga new domain, dynamics the studentsprogressively develop a keener sense of the kinds of outcomes they can produce and begin to propose more specific actions that they feel will shed some light on the disease dynamics. Their "observationof facts and suggested meanings or ideas arise and develop in correspondencewith each other. The more facts of the case [that] come to light in consequence of being subjectedto observation,the clearer and more pertinentbecome the conceptions of the way the problemconstitutedby these facts is to be dealt with" (Dewey, 1938/1998, p. 173). Their suggestions become ideas that, when examined in reference to the

UNDERSTANDINGTHROUGHDYNAMIC MODELING

487

situation,engenderthe capacityto predictand test solutions to their problematic situation. This section describesthe students'experimental design andexecutionin a parsimulation. Just as the their of ticipatory experimentalstateduringthe descriptions data collection phase were about their own state, their experimentaldesign involves varying their own behavioralpatternsto elucidatethe viral dynamics. Students offered ideas abouthow they could use variationsin their own behaviorto discover patternsin the viral behavior.As the experimentingproceeded and hypotheses were refined, the studentsimprovedtheir ability to predictthe viral outcome based on a certain set of (experimentallyconfigured)behaviors. Through experimentingand collecting additionaldataaboutthe relationshipbetween their own behaviorandthe behaviorof the virus (by conductingadditionalsimulations), they were eventuallyable to statethe rulesthatgovernthe viralbehavior.This process is a formof scientific experimentation, in which a system is probedundervarious conditions to reveal the underlying processes that govern the system's behavior. DuringEpisode 2, Rick proposedan experimentto figure out why some people didn't get sick when he exclaimed, "We should all meet each other."At thattime, his proposalwas ignoredby most of his classmates, as there was no community agreementon what aspect of the problematicsituationwas underinvestigation.As their investigationmoved forward,students'propositionsfor a variety of experiments to reveal the underlyingdynamics of viral transmissionbecame more frequent and focused. Episode 4 Researcher: Allison: Student: Allison: Rick:

Doug: Rick: Doug:

Do you have a strategyto avoid that [the virus]? Stay away from people. But you don't know who. That's what makes it confusing. I know how we could get it, everyoneturnon thembadgesand just turn'em aroundandthen whoever has the uh, whoever's thing lights up first. How 'bout all the people, each one [has a] partner,and then only meet with one person and whoever gets sick. Everyoneturntheirbadge aroundso no one can communicate with them and whoever's thing turnsred first. But can't the host not get sick, like the personwho has the virus his buttonswon't get red but he could give it to someone else? Yeah, we could pick groups,like um, they communicatewith each other,they communicatewith two people andif they get sick then these are the people who have the virus.

488

COLELLA

Stacy:

Go aroundthe roomagainlike we did before andthen as soon as your thing turnscolor, like, yell, out, you know what I'm saying,when it turnscolor,tryto see who was the firstperson. And then we could record,like, who we sharedwith.

In this episode, a number of students described possible experimental protocols. Rick wanted everyone to avoid meeting other people to determinewhose Tag showed viral symptoms (lit up) first. He felt that his plan would help determine the identity of Patient Zero (the initial host). Doug was concerned that Rick's plan did not control for the possibility that Patient Zero could just be a carrier and never display the symptoms of the virus. Stacy wanted to run an unconstrainedsimulation and watch for the first appearanceof viral symptoms. Over time, many students proposed experiments, and the group decided which ones they wanted to conduct, often based on a comparison between the data that the experiment was expected to produce and the currentlyavailable facts. Because there is a high level of iterativity and flexibility in participatorysimulations, it was easy to accommodate as many experiments as the students wanted to run. The studentsin the study exhibiteda remarkable level of pride and ownership about their proposed experiments.All studentspossessed the ability to articulate experiments,which, after all, were really prescriptionsfor alteringtheir own behavior in a way thatthey felt would illuminatethe rules of the virus. Any student accould offer an experimentalsuggestion or directthe groupto take a particular tion and observe the results.It was up to the groupto determinewhose suggestion made the most sense, given the problemat hand. Episode 5 Allison: Rick: I thinkwe shouldjust turnours on and wait and see whoever gets sick first. (Leapingout of his chair)THAT WAS MY PLAN! You got that on tape, right, I said it first!

...conducting the experiment ...

Rick: Student: Allison:

We're supposedto chill. Allison you wanna exchange? No, we're not supposedto have anybody. Everybody's supposedto have zero. Is everybody supposedto have zero? Stacy: Researcher: That's what I thought. Rick: This is my experiment! Tom: Oh, I get it. We're tryingto see if anybodyturnsup red. Student: One minute. I think we should give it 10 minutes. Allison:

UNDERSTANDINGTHROUGHDYNAMIC MODELING

489

The studentsconceived the experimentsandretainedcompletecontrolover the experimentalruns, though the facilitatorcould aid studentsduringthose runs. This studentcontrolwas possible because of a uniqueattribute simulaof participatory tions-there was no simulationunless all of the student-agents createdone. If any class memberbecame marginalized,eitherbecause he was confused as to the natureof the experimentorbecausehe was tryingto subvertthe experimental process, the group pulled him back in.10Rerunninga simulationor conductingan experiment in this environmentnecessitatedthe participationof every student. Otherwise, it was as if the simulationwas only partiallyrunning,a situationthat would not yield useful results. Episode 6 Stacy: Allison: Rick: Allison: Doug: Rick: Allison:

Oh look, it's red. Just only beep her once and that's the only person you meet with is Stacy. Then we're all gonna end up with it! Why? No, 'cause we have to see who's immune. I'm not going to beep her. I don't want to beep her. You have to, or else the experimentwon't work.

Inboth Episodes5 and6, therewas communitynegotiationaboutthe design andexecution of the experiments.Studentscontinuedto offer ideas for new experiments andask for explanationsaboutwhy certainpropositionswere expectedto yield particularpieces of informationfromthe simulation.But once the groupbegan to collect data,studentsexertedpressureon one anotherto comply with the statedprotocols. The nature of the participatorysimulation ensured that all of the class membersworked together.In this way, participatory simulationsdiffer from collaborativeenvironmentsin which the facilitatormust keep all of the studentstowith gether.As Allison explainedto her classmate,Rick, "youhave to [participate us] or else the experimentwon't work."

of Scientific Progression Activity


The studentsin this pilot studywere firstchallengedto meet a lot of people without ofthe simto articulate a clearunderstanding catchingthevirusandthenencouraged ulation.The participatory simulationthatenabledtheir activitieswas a motivating learningenvironment.Studentsworkedtogetheras they figuredout whatwas hap'0SeeGranott (1998) for a discussionon definingthe size of, and subsequentlyanalyzing,the unit of collaboration.

490

COLELLA

microworld,the studentsneededto unpening in the simulation.As in a traditional its dynamicsand derstand theunderlyingrulesofthe simulation to fully comprehend final outcome. They helped each other gatherevidence, define the problem, and buildtheoriesaboutthe dynamicsofthe system.Finally,they designedandexecuted experimentsto test theirhypothesesabouttherulesoftheir simulationenvironment. The studentslearnedaboutthese rules not by masteringa specific symbolic repreinteracsentationofthem butby consideringandmodifyingtheirown interpersonal tions andobservingthe resultingviraldynamicsuntil they could reliablypredictan experimentaloutcome.Forinstance,at the end of the study,they could predictwho Zeroandhow long itwould takefor would orwouldnotget sick aftermeetingPatient an infected person to show symptomsof the virus. During this study, the studentsplayed a total of six virus games. Each simulation game took only a few min to play; however, studentstypically spent more time-up to 25 or 30 min-discussing each game and planningtheir strategyfor the next one. In the firstfew games, studentswere not inquiringinto a well-defined problem. Instead,their focus was on general observationand data collection. As they agreedon a few they gainedfurther experiencein the simulationenvironment, later to In the that wanted solve. specific problems games, they were more they to confirm or deny data as and collected systematic they designed experiments their hypotheses. An analysis of the episodes from the first participatorysimulation game revealed that instances of data collection and preliminarydata analysis were more frequentthaninstancesof experimentaldesign. As the studentstriedto make sense of the first game, therewas much discussion abouteach individual'sexperiencein the simulation.Therewas almost no focus on designing experimentsto elucidate the dynamics of the system. As a result of their lack of experimentalplanning, Game Two followed a very similarpatternof behaviorand appearedto yield little new informationaboutthe dynamicsof the system. In spite of this aimless appearThe evidence that the students ance, Games One and Two were not unimportant. gatheredand the experiences that they accruedbecame the foundationfor their more systematicapproach to problemdefinitionandexperimentaldesign in Game Three. Game Threetook place on the 2nd day of participatory Duringthis activities.11 game, the studentsagreed on a problem:figuring out how the virus spreadfrom studentto student.Then, they worked togetherto analyze the datathat they had collected. More focused experimentaldesign emergedduringthis game. The concurrentpursuitof gatheringnew facts anddesigningandrunningnew experiments continuedthroughthe next three games, increasingin the numberof occurrences per game, until the group could articulatethe underlyingrules of the simulation.
simulationshas shown thatallowing time for inde"Our experiencein this and otherparticipatory pendentreflectionresults in more proficientproblemdefinitionand experimentaldesign.

UNDERSTANDINGTHROUGHDYNAMIC MODELING

491

This patternof activityis consistentwith the characteristics of scientific inquiry thatwe aimedto facilitate.As describedby Hall (1996), "inquiry proceedsby a reflective interplaybetween selecting conditionsin a situationthat framea problem andconceiving of relatedactivitiesthatwill bringabouta solution"(p. 211). In the simulationsenvironment, studentsframedmultipleproblemsandexparticipatory ecuted experimentalactionsto discoverthe solutionsto those problems.Although this pilot study does not allow us to conclude that the participatorysimulation alone caused studentsto engage in inquiry,it does allow us to observe that in this environmentstudents are able to define a problem, inquire into its nature, and solve the problem.Ourhope is thatthis experiencewill be one of many in which the studentsbuild and practicethe skills of scientific inquiry.

DISCUSSION
Ourpreviouswork with the ThinkingTags (Borovoy et al., 1996, 1998) informed much of the early design of participatory simulations.We wantedto keep the techof the nological aspects activityunobtrusive,even thoughthe computerswere methe simulation. The simulationitself utilized the Tags' ability to structure diating in this case interactions, by handlingthe spreadof the virus. As we continuedour classroomwork with participatory simulations,we looked more closely at the specific ways thatactivitieslike this one can supportcollaborativescientific investigations. Overtime, a numberof design principlesfor constructingand implementing simulationshave emerged.Much work remainsto be done to deterparticipatory mine the best ways to use participatory simulationsand othersimilaractivities in the context of classroomlearninggoals. We hope thatthe following design principles will prove to be fruitfulstarting pointsas we continueto investigatethe educational efficacy of participatory simulations.

Keep the Technology Unobtrusive


Carewas takento preservenaturalsocial interactions,using the Tags to augment, not take over, communicationand collaboration.In the Participatory Simulations Study,this design choice accomplishedtwo important goals. First,the Tags do not between students.Second,thoughthe get in the way of the naturalcommunication the studentsbecome deeply engagedin the disease technology is quiteunobtrusive, experience. Therearemanywell-documentedandvariedexamplesof Computer-Supported CollaborativeLearning(e.g., Koschmann, 1996). Participatory simulationsprovide anotherexample of a computer-basedcollaborativeenvironmentthat fully use voice, gesture, supportsnaturalcommunicationamong students.Participants

492

COLELLA

and expressionto communicatewith one another,ratherthan sharinginformation throughtext and images on-screen. Students'interactionswith each otherand the simulationare not constrained by large monitorsor awkwardtechnology configurations.12 the minimal Moreover, technology display seems to encouragestudents to use their own imaginationand prior experience duringthe activities. The students are able to use social cues and knowledge abouteach otherto enhancetheir engagementin the game. PictureRick's pride when he exclaimed that he wasn't sick: "I'mthe man ... that'sright,I'm a clean headagain ... You all want to be like me." Orthe initial suspicion thatTom was the first carrier: "Whostartedout with it? I think Tom did. Why? Because ... look at him. (laughter)Sometimesyou can tell like that."Or the notion that Doug startedout with the virus because he was "dirty."On the last day of the project,two studentsrecall their experiences: Episode 7 Tony: You don't feel good when you have the virus unless there's somethingnot workingup there .... Yeah, 'cause I didn'tlike it, I got it [the virus] when I wasn't even in the room andthat was just upsettingto me. It's a hardthing to deal with. Say you have HIV or something,a virus,andit don't show up in your system rightaway, you could give it to someone else without knowing.

Episode 8 Doug:

The participatory simulationallows the inclusionof the priorknowledge,attitudes, habits,andintereststhatthe studentsbringto the experience.The studentswho parsimulationsdrawon the frameworkof the simulationand ticipate in participatory their own knowledge and imaginationas they experience in the simulation.They act andrespondas thoughthe simulationis real, even thoughthereis very little explicit visual supportfor the metaphorof the game. The Tags' minimaldisplay does not impairandmay supportthe students'ability or willingness to suspendtheirdisbelief aboutthe simulation,and the unobtrusive natureof the Tag technology supports rich interactions amonga largegroupof students. This result may have implications for designing engaging educational technology, the budget for which rarelyrivals that of pricey virtualreality games in which fancy graphicsandhead-mounted displaysprovideall of the context for a virtuallyreal experience.

12See Stewart,Bederson,andDruin(1999) andStewart, Bederson,andDruin(1998) for a Raybourn, differentapproachto enablingmultiple studentsto interactwith a single computer.

UNDERSTANDINGTHROUGHDYNAMIC MODELING

493

AddCoherent, World Consistent Rulesto the Experiential


For many years, role-playinggames have entertainedchildrenand adults. These to adoptcertainpersonas games, like Dungeons and Dragons, enable participants would in and in so doing requirethatthe participants behave like their characters the Internet,have expandedthe any given situation.Computers,and in particular range and popularityof these games (Turkle, 1995). The ThinkingTags create a new kind of role-playinggame thatcombinesthe immediacyof real-life adventure with the consistentrules of mediatedgames and microworlds.Withoutconstraining the communicationor the behaviorof the students,the Tags providea tremendous amountof structure in the environment. The Tags carrytheunderlyingrulesof the simulation(viralrules in this study)into the students'world.In some sense, the the studentsinto agentsin a microworld,even as they allow the stuTags transform dents to retaintheir own personalities. Bringing a microworldinto the realm of students' experience enables them to explore the underlying formal structureof that world without abandoningtheir own perspective.They makeuse of the consistentbehaviorof the Tags as they design experimentalprotocolsto reveal the rules thatgovern viral behavior.Each of these componentsof the participatory simulationarisesbecause the Tags createan environmentthatis initiallymysteriousbut,upon further reflectionand action,becomes transparent. The use of the Tags allows the studentsto reach transparency througha new path that drawsupon the students' own personal experiences and their own systematic explanationsof those experiences. When designingthis participatory simulation,we constructeda simulationthat thancreatinga simulationthatclosely mirexploresa few important concepts,rather rorsa real-lifesituation.Inthis example,we focus on the conceptsof latencyandimsimulationto HIV,we do munity.Thoughmanystudentscomparethisparticipatory not model any of the complexities of HIV transmission(and the infection rate of 100%is quitedifferentfromthatof HIV). We purposelyincludethe artifactof only being ableto meet another persononce in this simulationbecauseit makesthemodel tractable,not because it increases fidelity to a real-worlddisease. Roughgarden the most fundamental (1996) calledthis type of model, which seeks to capture parts of the system andillustratea generalprinciple,an idea model Participatory simulations allow us to createa richlearningenvironment thatis basedupona "smallcluster"of essentialideas (diSessa, 1986);in this case, latencyandimmunity.It may be thatmorecomplex systemsmodels arebetterexploredthroughothermedia,including microworldsand traditionalsimulationenvironments.

RecreateScientific in Interpersonal Phenomena Space


simulationsbringstudentsinto directcontactwith scientific phenomParticipatory enaby deployingthe phenomenain the students'own interpersonal space.Because

494

COLELLA

the simulationoccurs in real space with studentsas agents,thereis no gulf between participants'immediateexperience and the simulation-they are the population that is being affected. Participants' internalconditionsor responsesto the simulation are not treatedas separatefrom their inquiryinto the simulation.Thoughthe simulationis an intentionallycontrivedenvironment, the studentsare participatory compelled by their experience, often exclaiming duringthe virus simulationthat they have "died"or "caughtHIV." Farfrompreventinginquiryor impedingthe studyof important scientific matein to entry the simulation reduces the barrier rial,participants' personalexperience forthe design andexecutionof scientificexperiments.This reducedbarrier may occurbecause participatory simulationsask the studentsto considerand explorepatterns in theirown behavior.Any participant can collect data(by reportingon their own experienceorthatofa peer)orproposeanexperiment(by suggestinga new patternof humanbehavior).The virus simulationanalyzedhere supported a rich set of and in a context. outcomes experiential experimental socially meaningful

Facilitate Similar butNon-Identical Experiences


The activities in a participatory simulationare designed so thatevery studenthas a similarand meaningfulexperience.Similarexperiencesof the activity ensurethat the studentssharea commonbase fromwhich they explorethe simulation.When a describesher experienceof the activity, her classmatescan understand participant and relate to her description,in partbased on their own experiences. Later,when studentscollect dataandproposeexperiments, to take they areall equallyprepared partin these activities. Meaningfulexperiencesensurethatevery participant'sexwith respectto understanding the behaviorofthe whole simuperienceis important lation. Because a participatory simulationis a completely distributedsystem, no the whole simulation.No one student'sTag is more or less single Tag is "running" thanany otherstudent'sTag,3 and similarly,no student'sexperienceis important than any other student'sexperience. In fact, all of the any more or less important studentsmust contribute theirexperiencesto the groupdiscussionin orderto make it possible to understand the dynamicsof the system. Statedanotherway, each student's own vantagepointmustbe articulated andexploredin orderfor the groupto achieve an understanding of the whole system. The activities themselves enable a "social organizationin which all individuals have an opportunityto contribute something"and "to which all feel a responsibility"(Dewey, 1988, pp. 34-35).

'3Withthe possible exception of PatientZero who begins the infection;however, thatdesignationis chosen randomlyat the beginningof each game, meaningthat(in this case) PatientZero mightbe Doug the first game, Rick the second, and Liz the third,and so on.

UNDERSTANDING THROUGH DYNAMIC MODELING 495

However, not every experience is designed to be identical. Students whose Tags are immune to the virus have experiences that differ consistently from those of their classmates. Students who elect to behave in a particularmanner-perhaps meeting a lot of people or perhaps interacting with no one at all-also have incongruous experiences. The asymmetry of experience is created not by the differing talents of the students but by their differing experiences of the activity. In order to decipher the underlying mechanisms of the whole virus simulation, students must first develop an understandingof what happened to them and then listen to what happened to other people. As their descriptions build one by one, the students begin to develop an understanding of the system as a whole. The experiences that differ from the mainstreamcan then be identified as outliers, and alternative hypotheses can be proposed for those data points. simulationsenablea kindof collaborativelearningin which every Participatory child's experiencebuildstowardan understanding of the whole. However,participatory simulations set up a differentstructurefor collaborationthan many other forms of collaborativelearning do (Aronson, Blaney, Stephan,Sikes, & Snapp, 1978; Collins, Brown, & Newman, 1989; Slavin, 1996) because only a collaborative effort that engages all of the participants will enable the groupto constructa model of the whole simulation.Every studentneeds to sharehis or her experience of the simulationandevery studentmustparticipate in the experimental runsof the simulation. The process involved in building a collective understandingof the whole system pushes studentsto make theirthinkingovert (Brown & Campione, 1990) as they explaintheirideas andpredictionsto theirclassmates.This environment is particularlyrich for looking at the process of collaborationbecause the technology supports and mediates a problem context that involves the whole group, allows face-to-face collaboration,and provides a computationalsubstrate for experimentaldesign and execution.

EnableStudents to DeviseTheir OwnSolutions


In the pilot Participatory SimulationsStudy,studentswere not given a specific voto use when cabulary discussingthe rules of the simulation,norwere they given an alternative writtenrepresentation to describethe datathey collected orthe hypotheses they proposed.This lack of predefinedstructure for meaning-making activities to be both and appears promising problematic. Throughoutthe activities, but especially after the third game, the studentsendeavor to clearly express their ideas so that others can follow the points they are making. In time, they begin to agree on ways to talk aboutthe activity that everyone can understand. Here, they use the Tag numbers to express the concept of im-

munity.

496

COLELLA

Episode 9 Tony: Meredith: Tony: Meredith: Tony:

It was a patternlike that 20 21 thing. The numbers. It was 1 2 11 21. I said the 21 thing. It wasn't specific. It was specific-you knew what I was talking about. It was specific enough.

When Tony mentions the "20 21 thing,"he is referencingthe fact thathe thinks a certainset of Tags areimmuneto the virus.WhenMeredithcorrectshim by indicating exactly which Tags are immune,he protests,pointingout that even if his comment was not precise, it was sufficient for her to understand what he meant. This type of discourseis consistentwith thatof many otherparticipatory simulations we have run, in which participants digress from data collection or experimental design to settle on a precise meaning for "immunity" or "carrier." (In the case of immunity,studentstypically discuss whetheror not immunepeople can infect otherseven if they never show symptomsof the virus. In the case of a carrier, can ever show symptomsof the participants usually debatewhetheror not a carrier disease.) In its currentimplementation,facilitatorsof a simulationdo not provide correct definitions for these or other debated terms, even if the definition that the studentsultimately agree on is not precisely correct.This process allows the students to arrive at their own vocabulary for articulatingthe rules of the simulation. Although we have not yet undertakenextensive research in this area, the consistency with which various studentgroupswork to define specific meanings for their descriptionsof the simulation suggests that more researchinto this activity may be warranted. Similarly,the facilitatorin this studydid not suggest any kind of alternative representationfor the dataor the rules of the simulation.Some groupsof studentstry to design charts,diagrams,or othergraphicaldepictionsto aid in theiranalyses of the problem.The groupin this pilot studydrew a charton the boardof the last person that each studenthad met (duringEpisode 3). Unfortunately,unlike creating for systemwidebehaviorsand outcomes,it is quitedifficult to reprepresentations resent individual behaviors and outcomes in agent-basedsimulations (like this participatorysimulation) in a mannerthat illuminatesthe key interactions(e.g., Feigenbaum,Kannan,Vardi, & Viswanathan,in press). Other researchershave exploredthe cognitive gains thatpeople make when creatingtheirown representations (Bamberger,1998; diSessa, Hammer,Sherin,& Kolpakowski,1991; Greeno & Hall, 1997; Hall, 1996; Nemirovsky, 1994), and we hope to find a way to include such activities in futureparticipatory simulations. In fact, duringthis pilot study, the researcherfacilitatedall of the participatory simulationactivities. Althoughthis structure did enableus to explorethe students'

UNDERSTANDINGTHROUGHDYNAMIC MODELING

497

behavior and inquiry,it was not ideal for investigatingthe role of the facilitator. We expect that,as in problem-based learning,the facilitator'srole in participatory simulationsis complex and important to the educationalsuccess of the activity. A numberof teachersin high schools, universities,and graduateschools have begun simulationsin their classrooms. We anticipatelearnimplementingparticipatory ing more about the complex relationships that support this learning activity throughan analysis of theirclassroomexperiences. This pilot study suggests thatdeployingmicroworldsin real space offers an opportunityto reevaluatethe role thatstructured experiencescan play in understanding the mechanismsthat governpatternsandprocesses in the world. We hope that futureresearchwill help shed light on the ways that participatingin simulations can supportchildren's developing scientific understanding.

ACKNOWLEDGMENTS
This researchhas been generously supportedby the LEGO Group,the National Science Foundation(grants9358519-RED andCDA-9616444), andthe MITMedia Laboratory'sThings ThatThinkand Digital Life consortia. Special thanksto my advisor,MitchelResnick, for his supportandvaluableinthe project,to TimothyKoschmannfor his helpful commentsand sight throughout many enlighteningconversations,to Brian Smith for his candid feedbackand assistance, and to Janet Kolodner and Mark Guzdial for their encouragement.I would also like to thankJeremyRoschelle and an anonymousreviewer for their contributionsto the article. Thanks to the members of the Epistemology and LearningGroup at the MIT Media Laboratory,especially RichardBorovoy and Kwin Kramer.I am indebtedto the many studentsand teacherswho have participated in this project.

REFERENCES
andobjectconstruction: Two keys to learning.In Y. Kafai & Ackermann,E. (1996). Perspective-taking M. Resnick(Eds.), Constructionism inpractice:Designing, thinking,and learningin a digital world (pp. 25-35). Mahwah,NJ: LawrenceErlbaumAssociates, Inc. Aronson, E., Blaney, N., Stephan,C., Sikes, J., & Snapp, M. (1978). Thejigsaw classroom. Beverly Hills, CA: Sage. Bamberger,J. (1998). Action knowledge and symbolic knowledge: The computeras mediator.In D. Sch6n, B. Sanyal, & W. Mitchell (Eds.), High technology in low-income communities (pp. 235-261). Cambridge,MA: MIT Press. Booth Sweeney, L., & Meadows, D. (1996). Thesystems thinking playbook: Exercises to stretch and buildlearningandsystems thinking for Interactive NH: Laboratory Learning. capabilities. Durham, Borovoy, R., Martin,F., Resnick, M., & Silverman,B. (1998, April). GroupWear:Nametags that tell about relationships.Paperpresentedat Computer-Human Interaction(CHI), Los Angeles, CA.

498

COLELLA

augBorovoy, R., McDonald,M., Martin,F., & Resnick,M. (1996). Thingsthatblink:Computationally mented name tags. IBMSystemsJournal, 35, 488-495. N. (1997). Inventingkindergarten. New York: HarryN. Abrams. Brosterman, Brown, A., & Campione,J. (1990). Interactivelearningenvironmentsand the teaching of science and mathematics.In M. Gardner, J. Greeno,F. Reif, A. Schoenfeld,A. diSessa, & E. Stage (Eds.), ToAswarda scientificpractice ofscience education(pp. 111-140). Hillsdale,NJ:LawrenceErlbaum sociates, Inc. SupportedCollaboBruckman,A. (1998). Communitysupportfor constructionist learning.Computer rative Work:TheJournal of CollaborativeComputing,7, 47-86. Colella, V. (1998). Participatory simulations: Building collaborative understanding through immersivedynamicmodeling.Unpublishedmaster's thesis, MIT, Cambridge,MA. Colella, V., Borovoy, R., & Resnick, M. (1998). Participatorysimulations: Using computationalobjects to learn about dynamicsystems. Paperpresentedat CHI, Los Angeles, CA. Collins, A., Brown,J. S., & Newman, S. E. (1989). Cognitiveapprenticeship: Teachingthe craftof readHillsdale, ing, writing,andmathematics.InL. B. Resnick(Ed.),Knowing,learning,and instruction. NJ: LawrenceErlbaumAssociates, Inc. Dewey, J. (1938/1998). The patternof inquiry.In L. Hickman& T. Alexander (Eds.), The essential Dewey: Vol.2. Ethics, logic, psychology (pp. 169-179). Bloomington:IndianaUniversity Press. Dewey, J. (1988). Experience and education. In J. Boydston (Ed.), John Dewey: The later works, SouthernIllinois University Press. 1925-1953 (Vol. 13, pp. 1-62). Carbondale: diSessa, A. (1986). Artificialworlds and real experience.InstructionalScience, 14, 207-227. in the comdiSessa, A. (1988). Knowledge in pieces. In G. Forman& P. Pufall (Eds.), Constructivism puter age (pp. 49-70). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. Meta-representadiSessa, A., Hammer,D., Sherin,B., & Kolpakowski,T. (1991). Inventinggraphing: tional expertise in children.TheJournal ofMathematicalBehavior, 10, 117-160. Eisenberg,M., & Eisenberg,A. N. (1998). Shop class for the next millennium:Educationthroughcomhandicrafts. Journal of InteractiveMedia in Education,98, 1-25. puter-enriched Epstein,J., & Axtell, R. (1996). Growingartificialsocieties: Social sciencefrom the bottomup. Washington, DC: Brookings Institute. Eylon, B., Ronen, M., & Ganiel, U. (1996). Computersimulationsas tools for teaching and learning: in optics.JournalofScience Educationand Technology,5, 93-110. Using a simulationenvironment Feigenbaum,J., Kannan,S., Vardi,M., & Viswanathan,M. (1999). Complexityof graphproblemsrepresentedby OBDDs. Chicago Journal of TheoreticalComputerScience, 1999(5), 1-27. Fischer, K. (1980). A theory of cognitive development:The controland constructionof hierarchiesof skills. Psychological Review, 87, 477-531. Goldman, S. (1996). Mediating microworlds:Collaborationon high school science activities. In T. Koschmann(Ed.), CSCL:Theoryandpractice ofan emergingparadigm(pp.45-82). Mahwah,NJ: LawrenceErlbaumAssociates, Inc. Fromthe individual'sknowledgeto the ensembleprocess. N. (1998). Unit of analysisin transit: Granott, Mind, Culture,and Activity,5, 42-66. Greeno, J., & Hall, R. (1997). Practicingrepresentation: Learning with and about representational forms. Phi Delta Kappan, 78, 361-367. of expeas sharedactivity:SituatedcognitionandDewey's cartography Hall, R. (1996). Representation rience. TheJournal of the LearningSciences, 5, 209-238. media. CommuIshii, H., Kobayashi,M., & Arita,K. (1994). Iterativedesign of seamless collaboration nication of the ACM, 37(8), 83-97. Ishii, H., & Ullmer, B. (1997). Tangiblebits: Towardsseamless interfacesbetweenpeople, bits, and atoms. Paperpresentedat CHI, Atlanta,GA. Jackson,S., Stratford,S., Krajcik,J., & Soloway, E. (1994). Making dynamicmodeling accessible to 4, 233-257. pre-college science students.InteractiveLearningEnvironments,

UNDERSTANDINGTHROUGHDYNAMIC MODELING

499

Keller, E. F. (1983). Afeelingfor the organism: Thelife and work of Barbara McClintock.San Francisco, CA: Freeman. Koschmann,T. (Ed.). (1996). CSCL: Theoryand practice of an emergingparadigm. Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. Lillard,P. (1972). Montessori:A modernapproach.New York: Schocken. available D. Fish Ltd software, Meadows, banks, (1986). [Computer http://www.unh.edu/ipssr/index.html/ipssr/lab/fishbank.html]. Durham, NH: Institute for Policy and Social Science Research. Montessori,M. (1912). TheMontessorimethod.New York: FrederickStokes. National Committeeon Science EducationStandards and Assessment, N. R. C. (1996). National science educationstandards.Washington,DC: National Academy Press. Nemirovsky, R. (1994). On ways of symbolizing:The case of Lauraand velocity sign. TheJournal of MathematicalBehavior, 3, 389-422. J. Greeno,F. Reif, A. Newman, D. (1990). Using social context for science teaching. In M. Gardner, Schoenfeld, A. diSessa, & E. Stage (Eds.), Towarda scientificpractice ofscience education (pp. 187-202). Hillsdale, NJ: LawrenceErlbaumAssociates, Inc. Norman,D. (1993). Thepower of representation.Thingsthat makeus smart:Defending humanattributes in the age of the machine.Reading, MA: Addison-Wesley. Papert,S. (1980). Mindstorms:Children,computers,andpowerful ideas. New York: Basic Books. Project2061, A.A.A.S. (1993). Benchmarksfor scientific literacy.Oxford,England:OxfordUniversity Press. Pufall, P. (1988). Function in Piaget's system: Some notes for constructorsof microworlds. In G. Forman & P. Pufall(Eds.), Constructivism in the computer age (pp. 15-35). Hillsdale,NJ:Lawrence ErlbaumAssociates, Inc. Resnick,M. (1994). Turtles,termites,and traffic jams: Explorationsin massivelyparallelmicroworlds. Cambridge,MA: MIT Press. K., & Silverman,B. (1998). Digital Resnick,M., Martin,F., Berg, R., Borovoy, R., Colella,V., Kramer, manipulatives:New toys to thinkwith. Paperpresentedat CHI, Los Angeles, CA. Resnick, M., & Wilensky, U. (1998). Diving into complexity:Developing probabilisticdecentralized thinkingthroughrole-playingactivities. TheJournal of the LearningSciences, 7, 153-172. simulato computer Roberts,N., Anderson,D., Deal, R., Garet,M., & Shaffer,W. (1983). Introduction tion: A system dynamicsmodelingapproach.Reading,MA: Addison-Wesley. Roschelle, J. (1996). Learningby collaborating:Convergentconceptual change. In T. Koschmann (Ed.), CSCL:Theoryandpractice ofan emergingparadigm(pp.209-248). Mahwah,NJ:Lawrence ErlbaumAssociates, Inc. Roschelle, J., & Teasley, S. (1995). The constructionof sharedknowledge in collaborativeproblem collaborativelearning(pp. 69-97). New York: solving. In C. O'Malley (Ed.), Computer-supported Springer-Verlag. Rothberg,M. S. S., & Awerbuch,T. (1994). Educationalsoftwarefor simulationrisk of HIV.Journalof Science Educationand Technology,3, 65-70. Roughgarden,J., Bergman,A., Shafir,S., & Taylor,C. (1996). Adaptive computationin ecology and evolution: A guide for futureresearch.In R. Belew & M. Mitchell (Eds.), Adaptiveindividualsin evolvingpopulations:Models and algorithms(ProceedingsVol. 26, SantaFe InstituteStudiesin the Science of Complexity,pp. 25-30). Reading,MA: Addison-Wesley. In S. Lajoie & S. Salomon,G. (1993). On the natureof pedagogictools: The case of the writingpartner. as cognitive tools (pp. 179-196). Hillsdale,NJ: LawrenceErlbaumAssoDerry(Eds.), Computers ciates, Inc. and development. A case studyof educational Schoenfeld,A. (1990). GRAPHER: technology,research, In M. Gardner, J. Greeno,F. Reif, A. Schoenfeld,A. diSessa, & E. Stage (Eds.), Towarda scientific Associates, Inc. practice ofscience education(pp. 281-300). Hillsdale,NJ: LawrenceErlbaum

500

COLELLA

Seidner,C. (1975). Teachingwith simulationsand games. In R. Dukes & C. Seidner(Eds.), Learning with simulationsand games (pp. 11-45). Beverly Hills, CA: Sage. Senge, P., Roberts,C., Ross, R., Smith,B., & Kleiner,A. (1994). Thefifthdisciplinefieldbook: Strategies and toolsfor buildinga learning organization.New York: CurrencyDoubleday. SimCity [Computersoftware].(1993). Orinda,CA: Maxis. SimLife [Computersoftware].(1992). Orinda,CA: Maxis. Slavin, R. (1996). Researchon cooperativelearningand achievement:Whatwe know, whatwe need to know. Contemporary EducationalPsychology, 21(4), 43-69. Starr,P. (1994, Spring).Seductionsof Sim: Policy as a simulationgame. TheAmericanProspect, 17, 19-29. Sterman,J. (1994). Learningin and aboutcomplex systems. SystemDynamicsReview, 10, 291-330. A modelforco-presentcollabStewart,J., Bederson,B., & Druin,A. (1999). Single displaygroupware.: oration. Paperpresentedat CHI, Philadelphia,PA. Stewart,J., Raybourn,E., Bederson,B., & Druin,A. (1998). Whentwo hands are betterthan one: Enhancingcollaborationusingsingle displaygroupware.Paperpresentedat CHI,Los Angeles, CA. roles of software-basedscaffolding and Tabak, I., & Reiser, B. (1997, December). Complementary teacher-student interactionsin inquirylearning.Paperpresentedat the ComputerSupportfor CollaborativeLearning(CSCL) Conference,Toronto,Canada. school: Lessonsfor today.New York:TeachersCollege Press. Tanner,L. (1997). Dewey's laboratory based labs: Educationalresearchand standards.Berlin, GerTinker,R. (Ed.). (1996). Microcomputer many: Springer-Verlag. and the humanspirit. New York: Simon & Schuster. Turkle,S. (1984). Thesecond self Computers Turkle,S. (1995). Life on the screen. New York: Simon & Schuster. Turkle,S., & Papert,S. (1992). Epistemologicalpluralismandthe revaluationof the concrete.Journal ofMathematical Behavior, 11, 3-33. Vygotsky, L. S. (Ed.). (1978). Mind in society: The developmentof higherpsychological processes. Cambridge,MA: Harvard University Press. White, B. (1993). ThinkerTools:Causalmodels, conceptualchange, and science education.Cognition and Instruction,10, 1-100. Wilensky, U. (1993). Connected mathematics:Building concrete relationships with mathematical knowledge.Unpublisheddoctoraldissertation,MIT Media Laboratory, Cambridge,MA. andtesting compuWilensky,U. (in press). ConnectedScience: Learningbiology throughconstructing tationaltheories-an embodiedmodeling approach. InterJournalof ComplexSystems. simulations:Network-based Wilensky,U., & Stroup,W. (1999). Learningthroughparticipatory design ResearchAssociafor systems learningin classrooms.Paperpresentedat the AmericanEducational tion (AERA) Annual Meeting, Montreal,Canada.

Вам также может понравиться