Humanistic Mathematics Network Journal

Issue 7 Article 8

4-1-1992

Epistemological Pluralism: Styles and VoicesWithin the Computer CultureSherry TurkleMasssachusetts Institute of Technology

Seymour PapertMasssachusetts Institute of Technology

Follow this and additional works at: http://scholarship.claremont.edu/hmnj

Part of the Mathematics Commons, Scholarship of Teaching and Learning Commons, Scienceand Mathematics Education Commons, and the Women's Studies Commons

This Article is brought to you for free and open access by the Journals at Claremont at Scholarship @ Claremont. It has been accepted for inclusion inHumanistic Mathematics Network Journal by an authorized administrator of Scholarship @ Claremont. For more information, please contactscholarship@cuc.claremont.edu.

Recommended CitationTurkle, Sherry and Papert, Seymour (1992) "Epistemological Pluralism: Styles and Voices Within the Computer Culture," HumanisticMathematics Network Journal: Iss. 7, Article 8.Available at: http://scholarship.claremont.edu/hmnj/vol1/iss7/8

EPISTEMOLOGICAL PLURALISM:STYLES AND VOICES WITHIN THE COMPUTER CULTURE

Sherry Turlde and Seymour Papert

I. EPISTEMOLOGICAL PLURALISM

The prevai ling image of the computerrepresents it as a logical machine and computerprogramming as a technical, mathematical activity.Both the popular and technical culture haveconstructed computation as the ultimateembodiment of the abstract and formal . Yet thecomputer's intellectual personality has anotherside: our researchfinds diversityin the practice ofcomputing that isdenied by its social construction.When we looked closely at progranuners in actionwe saw [annal and abstract approaches; but wealso saw highly successful programmers inrelationships with their material that are morereminiscent of a painter than a logician. They useconcrete and personal approaches to knowledgethat are far from the cultural stereotypes of formalmathernatics.!

The diversity of approaches to programmingsuggests that equal access to even the most basicelements of computation requires accepting thevalidityofmultiple ways ofknowing and thinking,an epistemological pluralism. Here we use theword epi stemology in a sense closer [Q Piaget'sthan rothephilosopher's.Z In the traditional usage.the goal of epistemology is to inquire into thenature of knowledge and the conditions of itsvalidity; and only one form of knowledge. thepropositional. is taken to be valid} The step takenby Piager in his definition of "episternologiegenerique" was to eschew inquiry into the "true"nature of knowledge in favor of a comparativestudy of the diverse nature of different kinds of

48

knowledge. in hiscase the kinds encountered inchildrenofdifferentages.We differ from Piaget on an importantpoint. however. Where he saw diverse forms ofknowledge in terms of stages [Q a finite end point offormal reason. we see different approaches toknowledge as styles . each equally valid on its ownterms.

The barriers to acknowledging such pluralismare great. historically rooted in domains that go farbeyond computation. The formal. propositional wayof knowing. has been recognized traditionally as astandard. canonical style. Indeed. philosophicalepistemology has generally taken it as synonymouswith knowledge. Where concrete approaches toknowledge have been recognized at all . it has mostoften been as inferior ways of knowing. the kinds ofknowing adopted by necessity by those who have notyet mastered the canonical style. Thus Jean Piagetrecognizes in young children ways of thinking that donot conform to the canon but which are too coherentand efficacious to be branded simply as " wrong: ' Hecasts children's concrete thinking as a stage in aprogression [Q a formal style .4 Similarly. Claude Levi­Strauss recognizes "bricolage," a "science of theconcrete:' but relegates it [Q primitive societies. amanifestation of the " savage mind."S

More recently. concrete ways of thinking havebeen recognized in contexts that arenot easilydismissedas inferior. Ethnographers of science studying thedaily life of the laboratory have found that scientificdiscoveries are made in a concrete. ad hoc fashion. andonly later recast into canonically acceptableformalisms.v Scientific biography reveals that Nobel

HMN Journal #7

laureates relate to their materials in the concrete andtact ile s ty le of Levi-Strau ss ' bricoleurs.?Psychologists investigating adults' mathematicalthinking f ind that they use an effective and down -to­earth style very different from the abstract andformal math they were taught at school.8 Feministscholars have documented the power of concrete,contextual reasoning in a wide range of domains.?

With such contributions has come a growingconvergence of intellectual commitments to arevaluation of the concrete; but in general, theethnographers, psychologists, and femini st scholarswho have contributed to this revaluation have notseen computation as relevant to their concern s. Herewe present evidence which points toward s thepossi bility of new intellectual alliances.

In our research on progranuning style s. thecomputer has emerged as an important actor in therevaluation of the concrete, a privileged medium forthe growth of alternative voices in dealing with theworld of formal systems . The conventio nal routeinto formal systems, throu gh the manipulation ofabstract symbols. closesdoors that the computercanopen.The computer. with its graphics. its sounds, itstext and animation, can provide a pon of entry forpeople whose chief ways ofrelating to the world arethrough movement, intuition. and visua l impression.At the heart of the new possibilities for theappropriation offormal systems is the computationalobject. on the border between an abstract idea and aConcrete physical object. In the simplest case, acomp utational object such as an icon moving on acomputer scree n can be defined by the most formalof rules and is thus a mathematical construct. but atthe same time it is visible, almost tangible, andallows a sense of direct manipulation that only theencultured mathematician can feel in traditionalformal systems.Iv The computer has a theoreticalvocation : it can make the abstract concrete; it canbring formality down-to-earth.

We have studied computers and the culturesthat grow up around them in a wide variety of

HMN Journal #7

settings ranging from video game arcades toresearchlaboratories of artificial intelligence. In this paperwe draw parti cularly on a long-term line of researchon how people enter the culture of programming.Using clinical methods inspired by the Piagetian andpsychoanalytic traditions. we built up case studies ofchildren using computers in grade school settingsand college students taking a first progranuningcourse. We saw many manifestations of the concreteapproach, favored in our study by more women thanmen. We were also able to observe people reactingpoignantly to what they felt as a pressure to conformto an officially imposed style.U Although thecomputer as an expressive medium supportsepistemological pluralism, the computercultureofrendoes not. Our data points to discrimination in thecomputerculture that is determined not by rules thatkeep people out but by ways of thinking that makethem reluctant to loin in. Moreover. the existence ofdiverse styles of expert programming suppons theidea that there can be differen t but equal voices evenwhere the formal has traditionallyappeared as almostdefi nitionally supreme : in mathematics and thesciences.

Evelyn Fox Keller has remarked on thedifficultythat people face when they try to understand what itmight mean to do science in anything other than theformal and abstract canonical style. Describing sucha stylein the workofgeneticist Barbara McClintock,Keller notes that this is the " less accessible aspect"ofa scientist' s relationship to nature.12 In this essaywe describe people learning to program who arehaving experiences with formal systems that are inmany ways analogous to those of the bricoleurscientist or mathematician . One way the computercontributes to the revaluation ofconcreteapproachesin the domain of formal systems is by giving morepeople access to (and an experience of) them.

The computer force s general que stions aboutintellectual style to reveal an everyday face.U Evenschoolroom differences in how children programcomputers raise issues that come up in a moreabstract fonn in scholarly debates about scientific

49

objectivity. The computer makes ideas about non­canonical scientific voices more concrete andtherefore appropriable because we can relate themnot only to the science ofthe scientists but toour ownthinking .

Here we focus on descriptions of a concreteway of knowing; the formal. canonical style is wellknown and well defended. Yet, our discussion ofconcrete approaches is implicitly a discussion offormal ones; it contributes to the deconstruction ofthe canonical style as the only way to think. It alsosituates it: the supervaluationofrhe formal approachowes much of its strength within computation to thesuppon it gets in other intellectual domains. Fonnalthinking, defined as synonymous with logicalthinking, has been given a privileged Status whichcan be Challenged only by developing a respectfulunderstanding of other styles, where logic is seen asa powerful instrument of thought but not as the " lawof thought." In this view, " logic is 0;) tap not on top."As a carrier forpluralisric ideas about approaches toknowledge, the computer may hold the promise of

. catalyzing change not only within the computerculture but in the culture at large.

2. PERSONAL APPROPRIATION

Consider Lisa, eighteen, a first-year HarvardUniversity student in an introductory programmingcourse. Lisa fears that she will find the coursedifficult because she is a poet , "good with words notnumbers." In high school, she had always scornedteachers who had insisted that mathemat ics is alanguage . Yet, now, her first encounter with thecomputer has made Lisa ready to reconsider thisproposition and with it her characterization of herselfas someone "bad at math." Lisa starts well. surprisedto find herself easily in command of the coursematerial; but as the term progresses she reluctantlydecides that she "has to bea different kind ofpersonwith the machine." The pressure todo so is not fromthe computational medium. She says she can nolonger resist pressure from her teachers to think inways that are not her own.

50

Lisa wants to manipulate computer languagethe way she works with words as she writes a poem.There, she says , she"feels her way from one word toanother: ' sculpting the whole . When she writespoetry, Lisa experiences language as transparent,she knows where all the elements are at every pointin the development of her ideas. She wants herrelationship to computer language to be similarlytransparent. When she builds large programs sheprefers to write her own. smaller. building blockprocedures even though she could use prepackagedones from a program library; she resent s the opacityof prepackaged programs. Her teachers chide her,insisting that herdemand for transparency is makingher work more difficult; Lisa perseveres, insistingthat this is what it takes for her to feel comfonablewith computers .

Two months into the programming course,Lisa' s effons to succeed are no longer directedtowards trying to fee! comfonable.She has been toldthat the "right way" to do things is to control aprogram throu gh planning and black-boxing. thtechnique that lets you exploit opacity to piasomething large without knowing in advance hothe detail s will be managed . Lisa recognizes thevalue of these techniques - for someone else. Shestruggle s against using them as the starting pointsfor her learning. Lisa ends up abandoning the fight,doing things"theirway," and accepting the inevitablealienation from her work. She call s her efforts tobecome "another kind of person with the machine"her "not-me strategy:' and begins to insist that thecomputer is "ju st a tool." «It' s nothing much," shesays,"j ust a tool." Lisa ' s growing sense ofalienationdoes not stem from an inability to cope wit~

programming but from her ability to handle it in away that comes into conflict with the computerculture she has entered .

Aclassmate. Robin, is a pianist. Robinexplainsthat she masters her music by perfecting the smallest"little bits of pieces" and then building up. Shecannot progress until she understands the details ofeac h small pan. Robin is happiest when she uses thiS

HMNJoumalf

tried and true method with the computer, playingwith small computational elements as though theywere notes or musical phrases. Like Lisa. she isfrustrated with using prepackaged programs. She,tOO, has been told her way is wrong: "I told myteaching fellow 1 wanted to take it all apart and helaughed. at me. He said it was a waste of rime. thatyou shouldjust black box, thaI you shouldn 'tconfuseyourself with what was going on at that low level."

Lisaand Robincame 10 iheprogrammingcoursewith anxieties about not belonging (fearing that thecomputer belonged to male hackers who lived in "aworld apart") and their experiences in it only servedto validate their fears. 14 Although carefullydesigned and imaginative. the Harvard Universitycourse taught that there is only one right way toapproach thecomputer,3 waythatemphasizes controlthrough- structure and planning. There are manyvirtues to this computational approach (it makessensewhendividingthelaboronalargeprogrammingproject. for instance) but Lisa and Robin haveintellectual styles at odds with it. Lisa says she has" turned herself into a different kind of person" inorder to perform and Robin says she has learned to"fake it" Although both women got good grades inthis programming course, both have had to denywho they are in order to succeed.

Lisa and Robin's experiences make it clear thatthe computer can be a partner in a great diversity ofrelationships, that the computer is an expressivemedium that different people can make their own intheir own way . Yet those who wish to approach thecomputer in a non-canonical way are discouragedby the dominant computer culture, eloquentlyexpressed in the ideology of the Harvard UniversityCOurse. They are asked to change their style to suitthe fashion when they begin to interact with theofficial computer world, committed to a formal,rule-driven, hierarchical approach 10programming.1S

Like Lisa and Robin, their exclusion from thecomputer culture is perpetuated not by rules thatkeep them out, but by ways of thinking that makethem reluctant to join in. They are not computer

HMN Journal #7

phobic. They do not need to stay away because offear or panic; but they are computer reticent, Theywant to stay away because the computer has come [0

symbolize an alien way of thinking. They learn toget by and to keep a certain distance. One of itsmanifestations is the way they neutralize thecomputerthrough language wbich denies the possibility ofusing it creatively (recall how Lisa dismisses it as"just a tool").

In this way, discrimination in the computerculture takes the form of discrimination againstapproaches to knowledge, most strikingly againstthe one preferred by Lisa and Robin, an approach wecall bricolage.

3. BRICOLAGE

Levi-Strauss used the term bricolagetocontrastthe analytic methodology of Western science withwhat he called a "scienceoftheconcrete"in primitivesocieties.Jv'Ibe bricoleurs he describes do not moveabstractly and hierarchically from axiom to theoremto corollary. Bricoleurs construct theories byarranging and rearranging, by negotiating andrenegotiating with a set of well known materials.

Lev i-Strauss' descriptions of the two scientificapproaches, divested of his efforts t~ localize themculturally, suggest the variety of ways that peopleapproach computers. For some people in our study,what is exciting about computers is working withina rule-driven system that can be mastered in a top­down, divide-and-conquer way This is the" planner' s" approach taught in the Harvardprogramming course. This approach decrees that theright way to solve a programming problem is todissect it into separate parts and design a set ofmodular solutions that will fit the parts into anintended whole. Some programmers work this waybecause their teachers or employers insist that theydo . For others, it is a preferred approach; to them, itseems natural to make a plan, divide the task, usemodules and subprocedures.

51

Lisa, Robin, and others like them in our studyoffer examples ofa very different style. They are notdrawn to structured programming; their work at thecomputer is marked by a desire to play with theelements of the program, to move them aroundalmost as though they were material elements - thewords in a sentence, the notes in a musicalcomposition, the elements of a collage.

The bricoleurresembles the painter who standsback between brushsrrokes.Iooks at the canvas, andonly after this contemplation, decides what to donext. For planners, mistakes are missteps; forbricoleurs they are the essence of a navigation bymid-course corrections. For planners, a program isan instrument for premeditated control; bricoleurshave goals, but set out to realize them in the spirit ofa collaborative venture with the machine. Forplanners, getting a program to work is like "sayingone's piece"; for bricoleurs it is more like aconversation than a monologue. In cooking, thiswould bethe style of those who do not follow recipesand instead make a series of decisions according totaste. While hierarchy and abstraction are valued bythe structured programmers' planner's aesthetic,bricoleurprogrammers,likeLevi-Strauss' bricoleurs,prefer negotiation and rearrangement of theirmaterials.

For instance, Alex, nine years old, is a classicbricoleur. He attends the Hennigan ElementarySchool in Boston, the scene of an experiment inusing computers across the curriculum. There,studentsworkwith Logoprogramming andcomputercontrolled Lego construction materials .The work isboth frequent enough (at least an hour a day) andopen-ended enough for differences in styles toemerge.

When working with Lego materials andmotors,most children make something move by attachingwheels to amotor that makes them tum. They see thewheels and motor through abstract concepts ofwhatthey are for: the wheels roll, the motor rums. Alexgoes a different route . He looks at the objects

52

concretely, without the filter of abstractions. Heturns the Lego wheels on their sides to make flatshoes for his robot and harnesses one of the motor'smost tangible features: thefact thatit vibrates. Whena machine vibrates it tends to travel, somethingnonnaUy to be avoided. When Alex runs into thisphenomenon, his response is to make his robot(stabilized byits flat"wheel shoes") vibrate and thusmove forward. When Alex programs in Logo helikes to keep things similarly concrete.

Learners are usually introduced to Logoprogramming through the "turtle," an icon on acomputer screen that can be commanded (Q movearound the screen and leave a trace as it goes. So, forexample, the turtle can be told to move forward ahundred steps and turn ninety degrees with thecommands FORWARD 100 RIGHT 90. Four suchcommands would have the tunle drawing a square.Programming occurs when a set of commands suchas REPEAT 4 [FORWARD 100 RIGHT 901. aredefined asa procedure:TO SQUARE. Alternatively.a suhprocedure TO SIDE might be defined andrepeated four times.

Alex wants to draw a skeleton. Structuredprogramming views a computer program as ahierarchical sequence. Thus a structured programTO DRAW SKELETON might bemade up of foursuhprocedures: TO HEAD, TO BODY.TO ARMS.TO LEGS. just as TO SQUARE could be built upfrom repetitions of a subprocedure TO SIDE. Alexrebels against dividing his skeleton program intosubprocedures; his program inserts bones one byone, marking the place for insertion with repetitions

.. of instructions . One of the reasons often given forusing subprocedures is economy in the number ofinstructions. Alex explains that doing it his way was"worth theextra typing" becausethe phraserepetitiongave him a" better sense of where Iam in the pattern"of the program. He had considered the structuredapproach, but prefers his own style for aestheticreasons: " It has rhythm," he says. In his opinion,using subprocedures for parts of the skeleton is tOOarbitrary , preemptive, and abstract. "It makes you

HMN Journal #7

decide how to divide up the body and perhaps youwould change your mind abou t what goes togetherwith what. Like, I might decide to think about thetwo hands together instead of each hand with thearms! '17

In his own way. Alex has resisted the pressureto believe the general superior to the specific or theabstract superior to theconcrete. For Alex. thinkingabout handsas a subset of arms is toofar away fromthe reality of real hands.just as taking a motor thatwas most striking as a vibrating machine and usingit to tum wheels in the standard fashion was too faraway from the real motor he had before him. Whilethe structured programmer starts with a clear plandefined in abstract terms, Alex lets the productemerge through a negotiation between himself andhis material.

Anne, also nine years old. is another bricoleurprogrammer. Her favorite hobby is pain ting and shehas become expert at using sprites in programs thatproduce striking visual effects.IS A sprite is asecond Logo icon, a turtle that can be set in motion.Once you give a sprite a speed and a heading, itmoves with that state of uniform motion untilsomething is done to change it, just like an objectobeying Newton's first law.

In one of Anne's programs, a flock of birds(each bird built with a sprite) flies through the sky,disappears over the horizon, and reappears someother place and time. Ifall the birds were red. then itwould be ea sy to make themdisappearand reappear.The command SETCOLOR :INV ISIBLE wou ld getrid of them and SETCOLOR :RED would makethem reappear. Yet Anne wants the birds to havedifferent colors, and so making the birds reappearwith their original color is more complicated.

One method for achieving this end calls for analgebraic style of thinking: you make the programStore each bird's original color as the value of aVariable, then you change all colors 10 invi sible andrecall the appropriate variable when the bird is to

HMN Journal #7

reappear. Anne knows how 10 use this algorithmicmethod, but prefers one thai allows her to turnprogramming into the manipulation of familiarobjects. As Anne programs, she uses analogies withttaditional an materials. When you want to hidesomething on a canvas, you paint itout, you cover itover with something that looks like the background.Anne uses this technique to solve her programmingproblem. She lets each bird keep its color, but shemakes her program hide it by placing a screen overit. Anne designs a sprite that will screen the birdwhen she doesn 'I want it seen, a sky-colored screenthat makes the bird disappear. Anne is programminga computer, but she is thinking like a painter.

Thinking like a painter does not prevent Annefrom contributing a sign ificant technical innovationto her founh grade computer class. She is familiarwith the ideaofusing two sprites to form acompoundobject. Her classmates and teachers have alwaysdone this by putting the sprites side by side. Anne'sprogram is like theirs in using two sprites, one for thescreen, one for the bird, but she places the spri tes ontop of each other so that they occupy the same space.Instead of thinking ofcompound objects as a way ofgettinga picture to be bigger, she thinks ofcompoundobjects asa way ofgetting sprites toexhibita greatercomplexity of behavior, an altogether more subtleconcept.

Thus, Anne's level of technical expenise is asdazzling in its manipulation of ideas as in its visualeffects. She has become familiar with the idea ofdata structures by inventing a new one - her screenedbird. She has learned her way around a set ofmathematical ideas through manipulating angles,shapes, rates, and coordinates in her program. As abricoleur, her path into this technical knowledge isnot through structura l design , but through thepleasures of letting effects emerge.

As in the case of Alex, Anne does not write herprogram in "sections" that are assembled into aproduct. She makes a simple working program andshape s it gradually by successive modifications. She

53

starts with a single black bird. She makes it fly. Shegives it color. Each step is a small modification to aworking program that she has in hand . If a changedoes not work , she undoes it with another smallchange. She " sculpts." At each stage of the process,she has a fully working program, not a part but aversion of the final product.

Anne is perfectly ca pable of prod ucing aprogram with well delineated subprocedures,although her way of creating them has little incommon with the planner' s approach .tv Devoteesof structured programming would frown on Anne 'sstyle. From their point of view, Anne should designa computational objec t (for example, her bird) withall the required qualities built into it. She shouldspecify, in advance, what signals will cause her birdto change color, disappear, reappear, and fly . Onecou ld then forget about "how the bird works"; itwould be a black box. Anne's work dramatizes thefeature of bricolage that was so salient for Lisa andRobin: the desire for transparency. Like Lisa andRobin, she enjoys keep ing open the possibility ofrenegotiating their exact fonn. This means stayi ngin touch with that fonn at all times. The bricoleu rs inour study tend to prefer the transparent style, plannersthe opaque, but the program 's authorship is a criticalvariable in this preference. Planners want to bringtheir own programs to a point where they can beblack-boxed and made opaque, while bricoleursprefer to keep them transparent; but when dealingwith programs made by others, the situation isreversed. Now, the bricoleurs are happy to get toknow a new object by interacting with it, learningabout it through its behavior the way you wouldlearn about a person, while (he planners usually findthis intolerable.The planners' moreanalytic approachdemands knowing how the program works beforeinteracting with it. They demand the assurance thatcomes from transparent understanding, fromdissection and demonstration.

Despite the dominant ideology of thecomputercul ture which privileges the structured, hierarchic al,planner's style, Anne's case makes it clear that the

54

difference between planners and bricoleurs is not inquali ty of prod uct, it is in the process of creating it.In describi ng bricoleur programmers, we have madeanalogies to cooks and painters. Bricoleurs are alsolike writers who don't use an outline but Start withone idea, associ ate to another, and find a connectionwith a third . In the end, an essay "grown" throughnegotiation and association is not necessarily anyless elegant or easy to read than one filled in from anoutline, just as the final program prod uced by abricoleur can be as elegant and organized as onewritten with (he top-down approach.

Do programmers graduate from bricolage whenthey develop greater expe rtise? Will Anne becomea structured programmer in junior high ? Ourobservation s suggest that with experience, bricoleursreap the benefits of their long exp lorations, so thatthey may appear more "decisive" and like plannerswhen they program on familiar terrain. Also, ofcourse, they get better at "faking it." Still, thenegotiating style resurfaces when they confrontsomething challenging or are asked torry somethingnew. Brico lage is a way to organize work. It is notastage in a progression to a superior form . Interviewswith computer scientists and their graduate studentsturned up highly skilled bricoleurs, most of themaware that their style was "countercultural. " Indeed,there is a culture of programming virtuosos, thehacker culture, that would recognize many elementsof the bricolage sty le as their own.

Withinfeminist scholarship there isa substantialbody of literature that challenges the notion thathuman reason best expresses itself within terms ofWestern male gende r norms.20 For example, CarolGilligan 's work on moral reasoningcalls into questionthe idea of one privileged, mature way of thinki ng.Gilligan 's description of two approaches to moralreasoning is analogous in to ourcontrast between theformal , canonical approach to program ming and theconcrete style of the bricoleur.2l In the first.justiceis like a mathematical principle: to solve a problemyou set up the right algorithm, the right black 00%,you crank the handle, and the answer comes OuLZ2

HMN Journal #1

In the second, a contextualized argument is like aconcrete argument, one needs to stay in touch withthe inner workings of the arguments, with therelationships and possibly shifting alliances of agroup of actors whose interests need to be negotiated .

Despite Anne' s high level of achievement,theori sts ofstructured programming would criticizeher style for the same kind of reasons that a stagetheorist of moral development would classify themost impressively articulate contextual thinker at alower intellectual level then his or her "formal"colleague. In both cases, criticism would center onthe fact that neither is prepared to take the final stepto abstraction. Gilligan challenges this standardhierarc hy; she uses her o bserva tions of moralreasoning through concrete situations to rejectLawre nce Kohlberg ' s stage theory with itsdeterminate end point to development, an end pointin abs tract, universal principles.23 If one branc h ofthe development ofmoral reasoning moves towardsthe primacy ofUjustice," of the formal and analytic,Gill igan insists on equal respect for a di fferentbra nch o f development which lead s to wa rdincreasingly sophisticated ways of thinking aboutmor ality in concrete terms of care throughrelationship and connection.

Gilligan is concerned with both morality andepistemology when she says: "the moral problem[for women] arises from conflicting responsibilitiesrather than from competing righ ts and requires forits resolution a mode of thinking that is contextualand narrative rather than formal and abstract."24Her language expresses a primary concern with thecharacter of the morality which, as she says, requiresa cenain mode of thinking. This emphasis on thecharacter of the morality (ra ther than the mode ofthinking) is even more marked in recent writin gwhere she redescribes Kohlberg's theory as beingabout only one side ofmoral rea soning. In this view,Kohlberg is talking about justice, thus leaving theother side of morality, namel y care, to her.25

HMN Journal #7

This compromise which splits off the COlUenJ

of moraljudgments from themode of thinking aboutthem blunts the force ofGilligan's observations as achaIJenge to something more general than moralreasoning ; but that challenge is central to ourargument. Kohlberg's theory of the developmentofmoral judgment mirrors Piaget' s theory of thedevelopment ofintelligenceperse.Both express thevalue-laden perspective on intellectual growth thathas dominated Western philosophy. Piaget sees aprogression from egocentric beginnings to a final,"formal stage" when propositional logic and thehypothetico-deductive methcdvliberate" intelligencefrom the need for concrete situations to mediatethinking.26 In this vision, mature thinking is abstractthinking. We disagree: for us, formal reasoning isnot a stage, but a style. Gilligan ' s materials on thecountercultural style of moral reasoning. like thecountercuItural style in programming, challengesthe existence of hierarchical stages: for althoughPiager would place the "concrete" Anne squarely inthe prefonnal stage, her level of achievementundennines his assumptions about the superiority ofthe analytic and formal,

Thus, observation of programmers at workcalls into question deeply entrenched assumptionabout the classification and value of different waysofknowing. Itprovides examples of the validity andpower of concrete thinking in situations that aretraditionally assumed to demand the abstract Itsupports a perspective which encourages lookingfor psychological and intellectual developmentwithin rather than beyond the concrete and suggeststhe need for closer investigation of the diversity ofways in which the mind can use objects to think withrather than the rules of logi c.

4. OBJECTS

Sooner or later in building objects with Lege,students at the Henniga n School where we met Alexrun into the need for gears.27 Looking at their work

55

provides a good example of alternate styles appliedto working with the same problem, formal styles thatuse rules and concrete styles that use objects.

The motors in the construction se t tum at a highspeed with low torque. A car built by attaching thesemotors directly to the wheels will go very fast, butwill be so underpowered that the slightest slope orobstruction will cause it to stall. The solution to theproblem with Lego cars is the same as that adoptedby designers of real cars : use gears . Yet in order touse them effectively, children need to understandsomething about gear ratios.

If a smaIl gear drives a larger gear, the largergear will tum more slowly and with greater torque.It is the relative and not the absolute size of the twogears that counts. But when we interview children ,we find that some of them reason as ifthe size of onlyone gear matters, as if they were followin g a set ofrules such as "large gears are slow and strong" and"small gears are fast and weak." With out the notionof relative size, suchrules fail. Other children, and inour study, predominantly the girls, are less articulateand more physical in their explanations. Theysquirm and twist their bodies as they try 10explainhow they figure things out: and they get the rightanswer.28

Theori sts who look at intellectual developmentas the acquisition of increasingly sophisticated ruleswould say that children run into problems if the rulesthey have built are not yet good enou gh.29 The ideaof "closenes s to objects" enables us to consider adifferent kind of theory . Our observations suggestthat the children who did so well did not have betterrules, but a tendency to see things in terms ofrelationships rather than properties, access to a styleof reason ing which allowed them to imaginethem selves " inside the sys tem." They used arelationship to the gears to help them think througha problem.

This "reasoning from within" may not beadequate for all problems about gears, but for the

56

kind of problem encountered by the children in ourproject, it was not only adequate, but much lessprone to the errors produced by a too-simple set ofrules. Relational thinking puts you at an advantage:you do not suffer disaster if the rule is not exactlyright.

We have defined bricolage as a style oforganizing work that invites descriptions such asnegotiationa1 rather than planned in advance, whatWarren McCulloch called "heterarchical" ratherthan hierarchical.30 The story of the children andtheir gears serves to introduce another characteristicdisplayed by many bricoleur programmers. We callthis characteristic pruximality or closeness to theobject. There is little distance between Anne and hercomputatio nal objects. Like the children, who"reasoned from within" with the gears, Annepsychologically places herself in the same space asthe sprites. She experiences her screens and birds astangible, sensuous, and tactile . She is down there, inwith the sprites, playing with them like objects in acollage. When she talks about them her gestureswith hand and body show her moving with andamong them. When she speaks of them she useslanguage such as"}move here." Theobjectrelationsschool of psychoanalysis focuses on the waydevelopment progresses by a process ofinternalization of the things and people of the world.They come to live within us; they become the objecrswith which we think.31 When psychoanalysts talkabout "objects" they usually mean people.32 Herewe extend the idea of internalized "objects to thinkwith" to the domain of everyday relationships withan ifacts. It is not enough to ask whether individuals"like" or "don ' t like" to program because that putsthe question on too high a level of generalization."Liking"toprogramdependsonforgingapersonallymeaningful relationship with a computational object.a relationship that "fits." In forging this relationship.(here are several dimensions of choice. People callchoose among computational objects. For example.in the version of Logo used by Anne there was achoice between sprites and turtles, Some prefer theturtle, its static nature, the fineness in the way ir

HMN Joumal II

draws. Forothers, these same qualities are reason s toreject -the turtle as constraining, even unpleasant.They prefer the sprites, which move with flash andspeed.

People can (and do) choose different ways ofapproaching the same object. Computational object s,like turtlesand sprites standon the boundary betweenthe physical and the abstract You can see them ,move them, put one on top of another. Yet. they aremathematicalconstructions. Canonical programmerstreat a sprite more like an abstract entity, a Newtonianpanicle, while bricoleur programmers treat it morelike a physical object. a dab of paint or a cardboardcut-our.

Computational objects offer a great deal tothose whose approach to knowledge requires a closerelationship to an object experienced as tactile andco ncrete. Some people are comfortable withmathematical exercises that manipul ate symbols onq uadri ll e-ru led paper. Fo r many others ,computationalobjects offera physical pathofaccessto the world of form al systems. For them, theambivalent nature of computational objects maymake possible a first access to mathematics.33

Feminist critics have related the standard notion ofscientific objectivity to the social construction ofgender:objectivity in the sense ofdistancing the selffrom the object of study is culturally constructed asmale, just as male is culturally constructed asdistanced and objective. From this point of view,Anne's proximal styJe iscounterculrural, reminiscentof Kell er's description of geneticist BarbaraMcClintock's intimate relationship to the objects ofher scientific study. For McClintock. the practice ofscience was essentially a conversation with hermaterials. The more she worked with neurosporachromosomes (so small that others had been unableto identify them), "the bigger [they) got, and when Iwas really working with them I wasn't outside, I wasdown there. J was pan of the system . J actually feltas if I were right down there and these were myfriends.... As you look at these things, they becomepan of you and you forget yourself ."34

HMN Journal #7

Alex and Anne relate to computational objectsmuch as McCli'ltock related to chromosomes asdoesa successful computer science graduate studentLorraine who explains how she uses "thinking aboutwhat the program feels like inside" to break throughdifficult problems."Forappearancessake," she wantsto "look like I'm doing what everyone else is doing,but I'm doing that with only a small pan of my mind.The rest of me is imagining what the componentsfeel like. It's like doing my pottery." This is in sharpcontrast to programmers in the structured, canonicalstyle who use their favorite device of black- boxingas a way to maintain distance. The idea of the blackbox, designed not to be touched. mediates betweenthe structured (planning) style of organizing workand their relationship to computational objects.Structured programmers are not among the sprites ,they act on the sprites.

The contemptuous comment of one founh­grade boy who overheard a classmate talking about"being a sprite'twhen he programscan beinterpretedfrom this point of view. "Th at' s baby talk," he said." I am not in the computer. I'm just making thingshappen there: ' The remark reflects an insistence onboundaries and the development of a world viewthat will fall easily into line with the canonical,objective science whose gender-b~ed meaningsKeller has delineated.

In our research we find a close relationshipbetween bricolage, a style of organizing work, andproximality, a style of relating to the objects ofwork.35 Our data is consisrenrwith a model ofstylesas clusters of characteristics in which bricolage andproximalityfonn the nucleusofonec1uster("concretethinking") and planning and distality the nucleus ofthe other ("fonnal thinking"). These clusters areideal types: our contention is not that the attributesin each cluster are exactly correlated but that eachhas internal coherency in the way that a stableculture is coherent.

57

So for example, closeness to objects tends tosupport a concrete style of rea soning, a preferencefor using objects to think with, rod a bias against theabstract formulae that maintain reason at a distancefrom its objects.Conversely, adistanced relationshipwith objects supports an analytic, rule and planoriented style. Our theoretical conjecture is thatdegree of closeness to objects has developmentalprimacy; it comes first. The child fonns either aproximal or distant relationship to the world ofthings. The tendency to use the abstract and analyticor concrete and negotiational style of thinkingfollows.

Althoughcloseness to objectsfavorscontextualand associarional styles of work, it does not excludethe possibility of using a hierarchical one. Planningis not always an expression of personal style. It canbe acquired as a skill, sometimes because it isneeded to get a panicular job done. sometimes as afacade to hide rather than express individuality.

Indeed, our data suggest that we may beunderestimating the degree of association betweenprcximality and bricolage. Some people adoptelements of the canonical style because they feel asocial pressure to do so. In order to attract lessnegative attention, Lisa said that she decided to be adifferent kind of person, i.e., more of a "planner,"Robin says she "fakes it" and forces herself to blackbox. Lorraine affects the discourse of a distancedstyle while something very different is going on inher head. Some bricoleurs respond to the dominantethos of the computer culture by entering into aninauthentic relationship with the computer. Thiscanlead to a paradoxical reaction: frustra ted bricoleursappear at first sight to beextremelyrigid " planners."Some turn to a "cookbook" approach - like whenin third grade, we were told to divide fractions byturning "t he second fraction" upside down. Whendenied a chance to do their "real thinking," they turntoroles thatdo not require them to think at all. Peoplelike Lisaand Robin"escape toconformity," areactionthat muffles the manifestation of their differentvoices in computing. Nevertheless, those voices are

58

there. Recall the graduate student Lorraine, whosays she tries "to look like I'm doing what everyon eelse is doing ," in order to preserve "appearances."Her style is hidden beneath her effons to fit in.

In our culture, the structured, plan-oriented,abstract thinkers do not only share a style butconstitute an epistemological elite. We have neverseen a case in which someone claims to have feltpressure to move away from the canonical style.Thus, since the phenomenon of" faking it" goes onlyin onedirection, we conclude that true occurrence ofthe bricoleur/proximal combination are even morecommon than our raw count.

Another attribute associated with proximalityand bricolage when working with computers is atendency to anthropomorphize, to refer to the systemas though it had human qualities . Thanthropomorphi zation extend s fro m thecomputational objects ("That sprite doesn't want todo what I tell it now") to the computer itself. Anne.for instance, has no doubt that computers havepsychologies:"they think," she says«butcan't really,have emotions." She believes, however, that thecomputer has preferences. "H e would like it if youdid a pretty program." When it come s to technicalthings, Anne assumes [he computer has an aesthetic"I don ' t knowif he would rather have the program bevery complicated or very simple." Anne knows thathe computer is just a machine, but she sees inonetheless as a male companion, if only a limited!one. Anthropomorphization, both of the computesystem and its pans, does not follow from lack 0

technical expertise. It is a stylistic preference.Jf

Very young children are in fact uncertainwhether computers should be counted as alive or notalive, and argue the question hotly, debating thecomputer's aliveness on the basis of its psychology,its intentions, consciousness. and feelings. By ageten, most are sure that the computer is not actuallyalive. However, at this point, some children, like:Anne, continue to behave with and talk about thecomputer as if it were sentient. They brag that it is

HMN Joumal l J

helping them or complain that it is not. In this, theyare not showing confusion about biology. They donot think that the computer is alive the wayan animalis, rather that it has a "kind of life," the kind of lifeappropriate to a computer: it thinks.37

Others have a very different reaction. Oncetheyarenc longerperplexedbywhetherthemachinemight actuallybe alive biologically, they shy awayfrom anthropomorphization. When they complainabout the computer, they do so in objective terms: itis too slow, itdoes not have enough memory. Talkingabout the computer usually means talking abouttechnical details.

Lise Motherwell, a researcher at the Henniganschool, did an intensive investigation of eight fifthgrade st~ents. Motherwell found she could describechildren' s stances towards the anthropomorphizationof the computer by distinguishing two styles:relational and environmental. Relational children,like those we are calling proximal thinkers, treat thecomputer as much like a person as they can, whileenvironmental children, analogous to those wedescribe as preferring a distanced approach, treat itlike a thing .

Once they have placed the computer in the not­alive category, the environmental children tend tosettle with relief into treating it as a thing . This helpsthem to appropriate it through a relati onship thatinvolves distance, objectivity, and control. Therelational children, once having settled the questionof biological alivenes s, get more comfortable withthe machine by making it an interactive panner. Inthe computer they have found something in thedomain of formal systems to which they can relatewith informality. Three out of the four girls inMotherwell's study were relational; three OUt of thefour boys environmenta1.38

InMotherwell'sstudy.asinthestudyofchildrenand gears, gender seems implicated in, but not adefinitive influence on , style, co nsis tent withObservationsofadult computer cultures where some

HMN Journal #7

men are alienated from the dominant engineeringstyle and many women work creatively within it.Again, as in our examples of Anne, Alex, Lisa,Robin, and Lorraine, the concrete style did not implya lower quality of work. Concrete, proximal gearbuilders didJUSt as well and in some cases benerthantheformal thinkers; childrenwhoanthropomorphizethe computer are no Jess technically sophisticatedthan those who do not. The degree of concrete,proximal, and anthropomorphic thinking does notreflect expertise but preferred approach toknowledge.

5. GENDER, CLOSENESS, AND CONFLICT

Several intellectual perspectives suggest tharwomen would feel more comfortable with arelational, interactive, and connected approach toobjects and men with a more distanced stance,planning, commanding, and imposing principles onthem.39 Indeed, we have found that many women dohave a preference for attachment and relationshipwithcomputers and computational objects as a meansof access to formal systems. Yet in our culturecomputers are associated with a construction ofscience that Stresses aggression, domination, andcompetition. The cultural construction of scienceleads to a conflict that considerably complicates ourstory of how women appropriate technology, In thecase of computation this conflict is particularlyacute. From its very foundations, science hasdefinedits way of knowing in a gender-based language.Francis Bacon 's image of the (male) scientistputtingthe (female) nature "on the rack," underscores theway objectivity has been constructed not only interms of the distance of the knower from nature, butin terms of an aggressive relationship towards it (orrather towards her). And from its very foundations,objectivity in science has been engaged with thelangua ge of power, not only over nature, but overpeople and organizations. Such associations havespread beyond professional scientific communities;aggression has become pan ofa widespread culturalunderstanding of what it means to behave in ascientific way. Its methods are expected to involve

59

"demolishing" an argument and "knocking it down"to size. Here the object of the blows is not a femalenature but a male scientific opponent If science isfirst a rape. it is then a due1.40

The traditional discourse of computation hasnot been exempt from these connotations. Programsand operating systems are "crashed" and "killed."We write this paper on a computer whose operatingsystem asks if it should "abon" an instruction itcannot "execute."ln our ethnographic studies of thesocial worlds that grow up around computing, wehave found that this is a style of discourse that fewwomen fail to note. Thus, women are too often facedwith the not necessarily conscious choice of puttingthemselves at odds either with the culturalassociations of the technology or with the culturalconstructions of being a woman .

When Lisa had a confrontation with herinstructor about the proper way to program thecomputer culture and its canonical epistemologywere represented by a person in authority withwhom she argued. In other cases the tension comesfrom fears of what people might think rather than aconfrontation with what someone actually thinks.Lorraine who programs by imagining "what thecomponents feel like" ends her description of herprogramming style by adding : " Keep thisanonymous. I know it sounds stupid:' In both thesereactions there is a tension between an individualand an outside agency; but the conflict is internalized:the computer culture alienates by putting one inconflict with oneself.

When Lisa first found herselfdoing well in herprogramming course, she found it "scary" becauseshe felt she needed to protect herselffrom the idea of"being a computer science type." In high school,Lisa saw young men around her turning to computersas a way to avoid people: "They took the computersand made a world apan." Lisa describes herself as"turning off' her natural abilities in mathematicsthat would have led her to the computer. "I didn 'tcare if I was good at it I wanted to work in worlds

60

where languages had moods and connected you withpeople ." Although Robin had gone through most ofher life as a musician practicing piano eight hours aday, she, too, had fears about "guys who establishedrelationships" with the computer. "To me. it soundsgross to talk about establishing a relationship withthe computer. I don't like establishing relationshipswith machines. Relationships are for people."

In the vehemence with which many womeninsist on the computer's neutrality, on its beingnothing more than a mere tool, there may besomething more subtle going on than a clash betweenculture and personal style- a clash between personalstyle and sense ofself. Many women may be fightingagainst having a close relationship to a computer.For some, like Lorraine, it could be because theywant to belong to the dominant computer culture.But for others, the experience of closeness to theobject is a source of conflict with themselves.

Lisa, like Anne. placed herself in the space ofthe computational objects she worked with and shetended to anthropomorphize, responding to thecomputer as though it had (at least) an intellectualpersonality. In Lisa's case , her own style came tooffend her because it had led her to what sheexperienced as a too-close relationship with amachine. When Lisa began programming she sawherselfas communicating with the computer, but themetaphor soon distressed her. "The computer isn't aliving being and when I think about communicatingwith it. well that 's wrong. There's a cenain amountof feeling involved in the idea of communication

..and I was looking for that from the computer." Shelooked for it and she frightened herself. ..It washorrible . I was becoming involved with a thing. Iidentified with how the computer was going throughthings."

In our research we find that women expressuch sentiments with particularurgency. We observethat a conflict fuels their convictions. In many cases.they are most comfonable with a style of thinking illwhich they get close to the objects of thought. Th<

HMN Joumal 11

computer offers them such objects; but the closerthey get to them the more anxious they feel. Oneremedy for their anxiety is denial. The more thesepeople become involved with the computer, themore they insist that it is only a neutral tool. Again,their assertion is belied by the vehemence withwhich it is expressed.

Lisa's conflict with her instructor would beresolved in principle by a greater tolerance for herway of thinking; but addressing internal conflictsabout being close to computers requires more thantolerance. It requires profound changes in the culturethat surrounds the computer. For instance. if thecomputer is a tool, and of course it is , is it more likea hammer or more like a harpsichord?

The musician Robin is not di stressed by herclose relationship with her piano which is also amachine. Lisa who finds attachment to the computer"unnatural" is not upset by her passion for thebeautiful, heavy antique ink pens with which shewrites . If Lisa had been in music schoo l, it is mostlikely that she. like Robin, would not experience asthreatening her sense of communicating with herinstrument or her emotional involvement with it.Music students live in a culture that over time hasslowly grown a language and models for closerelationships with music machines.The harpsichord,like the visual arti sts' pencils. brushes and paints, isa tool , and yet we understand that arti sts ' encounterswith these can (and indeed. will most probably) beclose , sensuous, and relational. Indeed , the bestartists will develop highly personal styles of workingwith them

The development of a new computer culturewould require more than environments where thereis perm ission to work with high ly personalapp roaches. It would requ ire a new socialconstruction of the computer, with a new set ofintellectual and emotional values more like thoseapplied to harpsichords than hammers.41 Sinceincreasingly computers are the tool s people use towrite, to design, to play with ideas and shapes and

HMN Journal #7

images, they should be addressed with a languagethat reflec ts the full range of human experiences andabilities. Changes in this direction would necessitatethe reconstruction ofourcultural ass umptions abou tformal logic as the "law of thought" This pointbrings us full circle to where we began. with theassertion that epistemological pluralism is anecessarycondition for a more inclusive computer culture.

7. ROADBLOCKS AND OPENINGS FORCHANGE

While the computer suppons epistemologicalpluralism, the computer culture has not. In someways, though, the computer culture is catching upwith the potential of the computer. In particular,si gnifi cant openings for change within thetechnological culture come from a new emphasis oncomputational objects which is making itself felt indomains as diverse as debates about which personalcomputers are the best and how to build anificialbrains.

Its simplest manifestation is the fashion forusing icons to control personal computers. In atraditional IBM-style computer system control isthrough typing instructions. In an iconic system, thesame effects are achieved by moving screen symbols.The current technology for the act of movingsomething ona screen falls shon of what thecomputerindustry expects to provide quite soon, but existingsystems. such as the Macintosh's "mouse" and touchsensitive screens already give a tactile sense thatrecall sAnne'sexperienceofprogrammingascollage.

Even superficial use of icons is enough totransform the perception of the computer by peoplewho are using it in computationally simple ways .For example. it is commonplace knowledge thatmanywriters who began to use compurersreluctantly,as a necessary evil, now find they have friendlierrelationship s mediated by the icon s, the mouse, andthe cozier appearance of a Macintosh. Althoughthese panicular warmer relationships do not involveprogramming, their influence may mean that the

61

next generation of people like Lisa and Robin willcome to programming courses with a different senseof who "owns" the computer.

A multiplicity oftechnical methods does not byitself lead to pluralism. It can simply lead tocompetition. to which the computer industry isscarcely averse, nor are those computer users whoseem to enjoy conversations that engage them inheated debate about the merits of their favoritesystem. It is only when we understand the computeras a projective screen for different approaches toknowledge that we can listen to these conversationsas a striving for pluralism. Different people arecomfonable with different systems. When peoplefight about the IBM versus the Macintosh, what theymay betrying to do is defend their intellectual styles.Yet, the debate is cast as an effort to prove the otherside wrong; as if it would be impossible to proveboth sides right.

A multiplicity of technical methods can alsolead to elitism. The Logo language allowed Anneand Alex to program in their own ways. but in manyeducational settings Logo is defined as the computerlanguage for children who have not reached the topstage in Piaget's hierarchy. In such settings even assophisticated a thinker as Anne or as creative athinker as Alex would get their liberty at the cost ofhaving their intellects defined as immature. Similarly,the very success ofthe Macintosh has often been castin terms that reflect the elitism of the dominantcomputer culture. The Macintosh iconic interfacehas been marketed as "the computer for the rest ofus:' with the implication that the rest of us needthings made simple and do not want to be botheredwith technical things.

As it happens, the popularity of icons may besettling this argument. a conclusion which has manyinterpretations. Thedesignersofcomputerinterfacesmight interpret it as final proof of the technicalsuperiority oficons. A psychologist might read it asa challenge to the concrete/formal split; perhapsmost people are concrete thinkers most of the time ,

62

and formal thinking is used for acceptability orprestige or functionality. Others might simply saythat icons are "easier." Individually. these positionsdo not support pluralism. What is important here iswhether the support for icons is part of a larger shifttowards an acceptance ofconcrete ways of thinking .

If the shift goes far it will be through theconnection of the icons to something deeper. aphilosophy of"object oriented programming.t'O Inthe traditional concept of a program, the unit ofthought is an instruction to the computer to dosomething. In object oriented programming the unitof thought is creating and modifying interactiveagents within a program for which the naturalmetaphors are biological and social rather thanalgebraic. The elements of the program interact aswould actors on a stage. This style of programmingis not only more congenial to those who favor'concrete approaches, but it puts an intellectual valueon a way of thinking that is resonant with their own.In principle it could undermine thecanonical positionin at least two ways: first. within the world ofprogramming through legitimating alternativemethods; second, in the larger intellectual culture,by supporting trends in cognitive theory thaichallenge the traditional canon.

Until recently. prevailing models of cognitivetheory have bolstered the commitment ofpsychologists and educators to the superiority ofalgorithmic and formal thinking. They were givensupport by the cognitive theorists most influential inthe computer world, the leaders of the artificialintelligence community. In the late 1970s and early1980s, the model of Al with the greatest visibilitywas [he rule-based "expert system" with its model ofmind as a structured information processor. Criticsof how computers influence the way we think citedthe information processing model as demonstratingthe instrumental reason and the lack of ambiguitrallegedly inherent in all computational thinkingabout intelligence.43 Artificial intelligence is not II

unitary enterprise, however. and recently, anothermodel has become increasingly prominent'"emergent A"I.44

HMN Joumal#7

Emergent AIdoes not suggest that the computerbe given rules to follow but tries to set up a systemof independent elements within a computer fromwhose interactionsintelligence isexpected toemerge.Its sustaining images are drawn not from the logicalbUI from thebiological and social Familiesofneuron­like entities or societies of anthropomorphizedsubminds and sub-subminds are in simultaneousinteraction from which mind-likeprocess is expectedto emerge. These models are sometimes theorized innotions of "mind as society," where negotiationalprocesses are placed at the heart of all thinking.Those who espouse and support such models are farmore inclined to find bricolage acceptable than areclassical Piagetians. What concerns us here, is notmaking value judgments about these trends in AI.just as we are not advocating a choice between theuse of icons and the use of textual instructions incomputer operating systems. What doe s concern usis that the new trends - icons, object-orientedprogramming. actor languages, society of mind,emergent AI · all create an intellectual climate in thecomputer world which undermines the idea thatfonnal methods are the only legitimate methods.

Thus. recent technological developments, ininterfaces, programming philosophy, and artificialintelligence, have created an opening forepistemological pluralism. We began by presentingthe notion ofepistemological pluralism by referenceto three streams of thought - feminism, ethnographyofscience,and psychology - which,althoughdifferenIin many ways, converge in reasserting the importanceof objects in thinking. We close by noting theopportunity for new alliances between them andcomputation.

Louis Althusser writes about psychoanalysislhattheimportantbreakthrough was not any particularstatement about the mind, but the step ofrecognizingthe unconscious as an object of study that defines anew theoretical enterprise.4S Psychology long hadconsidered the rational and the conscious as thequintessential mental activity; Sigmund Freud shiftedthe ground to the irrational and the unconscious. The

HMN Journal #7

unconscious was not given recognition as only animportant factor. but rather became an object ofscience in its own right Similarly, we imagine theemergence of a science of thought which wouldrecognize the concrete as its central object

There isevery reason to thinkthat the revaluationof the concrete will open the computer culture toaccepting the computer as an expressive mediumwhich encourages distinctive and varied styles ofusc.There is everyreason to think that this pluralisticcomputer culture could be more welcoming andnunuring to women and to men. Gilligan has saidthat women can protect the recognition "of thecontinuing imponance of attachment in humanlife:>46 The evidence from our research onprogramming styles leads us to conclude with ananalogous speculation. Feminist scholarship couldmake a crucial contribution to the (until now)predominantly male computerculture by promotingthe recognition ofthediverse ways that people thinkabout and appropriate formal systems and byencouraging the acceptance ofour profound humanconnection with our tools.

NOTES

1. Research reports that emphasize approach toprogramming or programming style in the sensewe are using it here include Seymour Papert,

'Andrea di Sessa. Sylvia Weir. and Daniel Watt."Final Report of the Brookline Logo Project,"Logo Memos 53 and 54. Massachusetts Instituteof Technology, Cambridge, MA, 1979; SherryTurkle, "Computer as Rorschach:' Society 17(December 1980): 15-22; Sherry Turkle. TheSecondSelf: Computers and The Human Spirit(New York: Simon and Schuster, 1984),especially Chaplet 3; Sylvia Weir, CultivatingMinds: A Logo Casebook (New York: Harperand Row. 1987); Sherry Turkle, Donald Schon,Brenda Nielsen, M. Stella Orsini. and WimOvermeer, "Project Athena at MIT:'(unpublished ms., May 1988); Lise Motherwell,

63

"Genderand Style Differences ina Logo-BasedEnvironment" (Ph.D. diss.• MassachusettsInstitute of Technology, January 1988); IditHare!, "SofrwareDesign for Learning: Children ' sConstruction of Meaning forFractionsand LogoProgramming," (Ph.D. diss ., MassachusettsInstitute of Technology, June 1988).

2. Piaget's useofthe tennepistemology is pervasivein his writing. See, in particular. Introduction aEpistemologie Cenetiquevol. 1-3(Paris:PressesUniversi taires de France, 1950). Alvin Goldmandiscusses the modem redefinitionof the fieldofepistemology as something as close topsychology and sociology as to philosophy inEpistemology and Cognition (Cambridge, MA:Harvard University Press, 1986).

3.Foracriticalandpolemicalaccountof this history,see PaulM.Churchland.A NeurocompuzationalPerspective: theNatureof MindandtheStructureof Science (Cambridge, MA : The MIT Press.1989).

4. See for example, Jean Piagetand Barbel Inhelder,The Growth of LogicalThinkingfrom Childhood/0 Adolescence (New York: Basic Books, 1958).

5. Claude Levi-Strauss, The Savage Mind (Chicago:University of Chicago Press, 1968) .

6. A sample of relevant studies in scientificethnography is provided by Karin Knorr-Cerinaand Michael Mulkay, eds., Science Observed.Perspectives on the Social Studies of Science(London: Sage Publication s, 1983). See alsoKarin Knorr-Cetina, The Ma nufa cture ofKnowledge: An Essay on the Constructivist andCo ntextual nature of Sci ence (Pergamon:Oxford, 1981 ); Bruno Latour and StephenWoolgar, Laboratory Lif e : Th e SocialConstruction ofScientific Facts (Beverly HillsCA: Sage, 1979); Sharon Traweek, Beamtimesand Lifetimes , The World of High EnergyPhysicisrs (Cambridge, MA:Harvard UniversityPress, 1989).

64

7. EvelynFox Keller, A Feeling for the Organism:The Life and Work ofBarbara McClintock (SanFrancisco : W.H. Freeman, 1983).

8. A sample of studies on everyday thinking iscontained in Barbara Rogoff and Jean Lave,eds., Everyday Cognition. Irs Development inSocial Context (Cambridge, MA: HarvardUniversity Press, 1984). Also see, Jean Lave,Cognition in Practice: Mind. Mathematics andCulture in Everyday Life (Cambri dge:Cambridge Universi ty Press, 1988).

9 . See, for example, Mary Field Belenky, BlytheMcV icker Clinchy, Nancy Rule Goldberger,and Jill Mauuck Tarule, Women's Ways ofKnowing: the Development of Self, Voice. andMind (New York: Basic Books, 1986). Editedcollections thatfocusonapproaches to knowingin science include: Ruth Bleir, ed., FeministApproaches to Science (New York: Pergamon,1986) and Sandra Harding and Merrill B.Hinti kka, eds., Discovering Reality. FeministPerspectives on Epistemolo gy, Metaphysics,Methodology , and Philosophy of Science(London: Reidel , 1983). An overview thathighlights many of the issues we deal with inthisessay is provided by Elizabeth Fee, "Critiques ofModem Science:TheRelationship ofFeminismto Other Radical Epistemologies" in Bleir,Feminist Approaches .

In this essay we situate ourposition by focusing ontwo writers, Carol Gilligan and Evelyn FoxKeller. Gilligan, with her emphasis on moraldiscoursemight seemoutofplace inadiscussionof non -canonical approaches to science andtechnology. Butherewe arguethat key issuesinthe critique of science are not about scientificreasoning but about reasoning. Juxtaposingmoral and computational reasoning helps US

make this point. In addition, Gilligan's criticalrelationshipto thetheories ofLawrence Kohlbertis analogous to our own critical relationship toPiaget's work. We emphasize Keller becauseherworkunderscoresasdoesours.theimponancc:

HMN Joumal l l

of relationships withobjects in the developmentof non-canonical styles. Using Gilligan andKeller as a contrasting pair allows us to highl ighttwo different dimensions of what we call theconcreteapproachto science. SeeCarolGilligan,In A Different Voice . Psychological Theory andWomen's Development (Camhridge: HarvandUniversity Press, 1982); Evelyn Fox Keller, AFeelingfor the Organism. The Lif e and Work ofBarbara McClintock; Evelyn Fox KellerRefle ctions on Gender and Science (NewHaven: Yale University Press, 1985).

10. See Philip J. Davis and Reuhen Hersh, TheMathematical Experience (Boston: HoughtonMifflin , 1981 ) and Seymour Papert, "TheMathematical Unconscious"inJudith Wechsler,ed. , Aesthetics and Science (Cambridge, MA :MIT Press, 1980).

Turkle, The Second Self: Computers and theHuman Spirit (New York:SimonandSchuster,1984).

14. Lisa and Robin were pan of a larger study ofHarvard and MIT students taking introductoryprogramming courses. The study found anxietyaboutanidentityas a"compmerperson" tobe animponan t aspectofreticencetow~computm,

especially among women. See Sherry Turkle,"Computational Reticence: Why Women FeartheIntimate Machine,"inCherts Kramarae, ed.,Technology and Women's Voices: Keep ing inTouch (New York: Pergamon, 1988). See alsoSara Kiesler, Lee Sproull, and Jacquelynne S.Eccl es, "Poolhall s, Chips, and War Games:Women in the Culture of Computing,"Psychology ofWomen Quanerly , 9 (1985) .

11. For grade school children we worked with fort ycases. Of the twenty girls in ourstudy, founeenpreferred concrete approaches, of twenty boysthere were four who followed this route. In thestudy of college students taking a firstprogramming course,ofthefifteen women, ninewere concrete styleprogrammers,oftifteenmen,four. Because of our interest in spontaneousapproaches we classified as concrete thinkerssome students who finally adopted elements ofthe canonical approach in order to please theirteachers. (See forexample, thecasesofLisa andRobin, below.) In ourresearch, themale/fonnaland female/concrete dichotomy was mostdramatic in a predominantly white, wealthyprivate school in the South where traditionalpatterns of socialization would favor boyslearning the ways of control, hierarchy, anddistance and girlslearningthewaysofnegotiationand closeness.

12. Keller, A Feeling for the Organism, 1980.

13. For a fuller discussion of the computer as anevocative and concretizing object see Sherry

HMN Journal #7

15. In 1987,Turkleconducted asurvey of37 womenmembers of a localcomputer society. Of these,17 reponed feeling pressure to change theirpreferred ways of working with thecomputer inorder to be more acceptable to the dominantcomputer culture. "I got my wrists slappedenough times and Ichangedmy ways,"said oneofthem.acollege student for whomprogrammingonherMacintosh wasaprivate passionuntil sheentered MIT.

16. Levi-Strauss, The Savage Mind . Levi-Strausscontrasted bricolage with Western science,ignoring the significant aspects of bricolagepresentin the latter. Several recent writers havewritten in a way that begins to redress thisimbalance. See for example, Paul Feyerabend.Against Method. The Outline of An AnarchisticTheoryofKnowledge (Lo ndon NLB, 1975); N.R.Hanson, Patterns of Discovery (Cambridge:Camhridge University Press, 1958 ); LudwigWingenstein,Philosophical Investigations (NewYork: MacMillan, 1953). In a less formal vein,see Richard Feynman, Surely You Must BeJoking Mr.Feynman (New York: Norton, 1985).

65

17. In its ideal. the structured method would have theprogrammer go beyond subprocedures to makeone procedure that could be given differentparameters to produce arms and legs, right andleft sides, even differently shaped people. Thisaesthetic, known as "procedural abstraction:'wants to see a right arm and a left leg disappearinto a generalized abstract idea of " limb." Butfor someone like Alex, the top priority is stayingin touch with the concrete. He is aware of theimponance oforga nizing hisprogram in ordertofind his way around it, but he does so by givingit what hecalls"rhythm" ratherthan a hierarchicalstructure of procedures and subprocedures.

18. Anne's program has the merit of showing incompact form a set of qualities characteri stic ofthe bricoleur. but usually more diffusel yrepresented . For a more detailed account , seeTurkle, The Second Self, pp. 110-15.

19. Bricolage does not exclude the use ofsubprocedures: it simply does not give their apriori delineation the status of a privilegedmethod. For example, a part of a holisticallyconceived program can be demarcated as asubprocedure at any stage of programming.Subprocedures need not be " black boxes"; theytoo can be developed as the program grows as awhole. Indeed, the bricole ur may use - ssubprocedures programs that happen to be"lyingaround," possibly even programs that wereoriginally made for very different purposes.

20. See for example the writings on scientificepistemology referred to in note 16.

21. Gilligan, In A Different Voice. For a criticaldisc ussion ofGilligan's proposals and her replysee Linda K. Kerber et. aI., "On In A DifferentVoice: AnInterdisciplinaryForum," Signs 11,no.2 (Winter 1986); 304-33. Its methodologicalcriticisms of Gilligan 's treatment of therelationship between "voice" and gender do not

66

detract from how her subjects illustrate the wayof thinking we call "bricolage.'

22. It is an eleven-year-old, Jake, who whenconfronted with a moral dilemma describes it as"sort of like a math problem with humans ." SeeGilligan, In A Different Voice , p . 26

23. Gilligan, In A Different Voice , 30ff. Kohlberghad already been challenged on other grounds.See,forexample,JohnGibbs," Kohlberg's Stagesof Moral judgment: A Constructive Critique,"Harvard EducationReview,47, no.4 (February,1977): 43-61. Similar issues have been raised incritiques of Jean Piaget, See, forexample,StevenToulmin,HumanUnderstanding (Princeton,NJ:Princeton University Press, 1972). Toulmi nargues that Piaget's experimental investigationsreflect an a priori commitment 10 a Kantianposition. We stngleoutToulmin, because unlikemost of Piaget ' s critics he does not quarrel withthe detail of how the stages are described butwith the epistemological assertion of [he finalend point.

24. Gilligan, In A Different Voice , p . 19.

25. Carole Gilligan and Jane Attanucci, "Two MoralOrientations," in Carol Gilligan, Janie VictoriaWard andJill McCleanTaylor,eds.,Mapping theMoral Domain (Cambridge: HarvardUniversity Press).

26. Piaget and Inhelder, The Growth of LogicalThinkingfrom Childhood to Adolescence.

27. These experiments with Lego and programmingare undenaken in a Piagetian spirit. See JeanPiaget, La Prise de Conscience (Paris: PressesUniversitaires de France, 1951) for experimentsthat deal with how mechanisms work. For apersonal statement about the power of gears asan introduction to formal systems, see SeymourPapert , "The Gears of My Childhood ," in

HMN Joumal#7

Mindstorms: Children, Computers, andPowerju1Ideas (New York: Basic Books, 1980).

28. Our sample does not allow us to say that girls didsystematically bener than boys. Research is inprogress on this point. Our present discussion isabout stylesofexplanation (rule-driven vs.body­syntonic) notdistribution of abilities.

29. For example most of those inspired by theCarnegie-Mellon schools of artificialintelligence. See, Ryszard S. Michalski et al,eds .• Machine Learning: An Artificiallmelligence Approacn (Los Altos, CA: MorganKaufmann, 1983).

30. Warren McCulloch, Embodiments of Mind(Cambridge, MA: MIT Press, 1988).

31 . For anexcellentoverviewof theobject relationsperspective perspective, see Jay R. Greenbergand Stephen A. Mitchell, Ob ObjectRelatiansinPsychoanalytic Theory (Cambridge: HarvardUniversity Press, 1983).

32. In contrast, D.W.Winnicott hassomesuggestiveideas about thepowerofthe"transitional object"- the baby's blanket, the teddy bear - that indevelopmental terms mediates betweenexperience of self and non-self. In the currentcontext. itsuggests thepower ofthe inanimate ininner life. See Playing and Realiry (New York:Basic Books, 197 1 ).

33. The Logo turtle was designed to be "bodysyntonic:'toallowusers topUIthemselves initsplace. When children learn (Q program in Logo.they are encouraged to work out their programsby"playing turtle." Theclassicexample of thisis developing the Logo program for drawing acircle. This is difficult if you search for it byanalytic means (you will need to findadifferentialequation), but easy if you put yourself in theturtle's placeand pace itout. (The turtle makesa circle by going forward a little and turning a

HMN Journal #7

little, going forward and little and turning a little,etc.) Tunles are a path into mathematics forpeople whose surest route is through the body.See Seymour Papert, Mindstorms.

34. Keller, A FeelingfortheOrganism.p. 117. Kellerdeseribes McClintock's appmach as dependenton acapacity to"forget herself,"immerse herselfin observation, and "hear what the.material hasto say." (page 198)

35. In the 70caseson which werepon here, 40 gradeschool children and 30 college students, wefound these two dimensions of approach toprogramming inall but 9 cases. Thus, empirically,we sometimes find eachaspect of theconcreteapproach - bricolage as a style of organizationand cJoseness to the object - without thepresence of the other. In particular, one findspeople who are planners but whoenjoya closerelationship with concrete objects (and whoexperience computational objects this way).Onthe pairing of planning and what they call aninteractive style with the computer, seeRosamund Sutherland and CeliaHoyles,"GenderPerspectives on Logo Programming in theMathematics Curriculum." inCelia Hoyles. eel.•Girls and Computers (London: Bedford WayPapersf34, Institute of Education, University ofLondon, 1988). .

36.Anthropomorphization is not limited to children.It is a habit of mind shared by adults, includingtechnical experts, When we say the computer"moves the queen" in a game of chess, theprogram has invited usto speak ofitasthough ithad intentions. And programs within acomputersystem interact with each other in a way thatsupportsmodelsof thecomputer ascomposed of«agents" incommunication.Computerscientiststalk about a concept such as recursion withanthropomorphic metaphors: one agent «callsup" another, "wakes up" another, and "passes onajob."Theysometimes evenrefer (Q theagentswithin acomputersystemascitizens ofa"society

67

of mind." See Marvin Minsky, Society 0/Mind(New York: Simon and Schuster, 1987) andPapert. Mindstorms.

37. Turkle, The Second Self, especially Chapter I.

38. Motherwell, "Gender and Style Differences inaLogo Based Environment." (see n.l )

39. Gilligan's work illuminates the relationshipbetween bricolage and gender and Keller speaksto the gender meanings of theproximal style ofrelating toobjects, betheyphysical objects suchas gears orchromosomes or conceptual objectssuch as the elements of programming. Apsychoanalyticperspectivewould place the rootsof such approaches at an early stage of childdevelopment. If in OUT culture , women are theprimary care takers forchildren. theearliestandmost compelling experiences of merging arewith the mother; differentiation and delineationtakeon gendermeanings.See forexample.NancyChodorow, The Reproduction 0/ Mothering:Psychoanalysis and the Sociology 0/ Gender(Berkeley : University of Califomia Press. 1978)and Keller.Reflectionson Gender and Science.Thetracesof suchearlyexperiencesarecuitural lyreinforced by continuing gender divisions ofparenting roles and by the very differentsocialization of men and women.

40. Keller, "BaconianScience:The Ans of Masteryand Obedience: ' in Reflections on Gender andScience , pp. 33ff; Carol yn Merchant,The DeathofNature (New York: Harper and Row, 1980);Donna Haraway. "The Biological Enterprise:Sex, Mind, and Profit From Human Engineeringto Sociobiology," Radical History Review 20:206-37.

41. On values for a new computer culture. seeSeymourPapert,"Technological Thinking versusComputer Criticism," Educational Researcher16, no. I (January 1987): 22-30.

68

42. For a non-technical discussion see, Alan Kay."Microelectronics and thePersonal Computer."SciemificAmerican 237 (September 1977): 230­44 and Alan Kay, "Software's Second Act,"Science 85 (November 1985): I22ff.

43. See, for example, Huben Dreyfus, Who,Computers Can't Do. The Limits 0/ Artijic/(JJIntelligence, 2nd ed . (Ne w Yolk: Harper anc!Row,1979)and Joseph Weizenbaum,ComputerPowerandHumanReason(SanFrancisco:W.HFreeman, 1976).

44.The reaction withinanificial intelligence againsiabstract,propositional. rule-dr iven meth ods wasgiven literary expression in the writings ofDouglas Hofstedrer . See for example "WakingUp fromthe BooleanDream,orSubcognition asCompu tation" in Douglas Hof sradter,MetamagicalThemas:Questingfor theEssenaof Mind and Pattern (New York: Basic Books.1985), pp. 631-65. Two other manifestations ofthis reaction are Marvin Minsky.SoderyofMinl.(Ne w York : Simon and Schuster, 1985) aneDavid E. Rumelhan, James L. McClelland, afIC

the PDP Research Group, Parallel DistribuusProcessing: Explorations in the Microstruauna/ Cognition (Cambridge MA : MIT Press, 198,For our more extended comments on the " 1'I\o~

Al s," see Seymour Papert, "One AIorMany" irDaedalus, 117, No. I (Winter (988): 1-13 aniSherry Turkle, "Artificial Intelligence anCPSYChoanalysis," in Daedalus. 117. No, I(Winter 1988): 24 1-68.

45. Louis Althusser,"Freud et Lacan," La NouvelkCritique Nos. 161-2 (December.January.Lw"65).

46. Carol Gilligan, In A Different Voice,p. 23­

[Puhlished in Signs: Journal of Women in Cui""and Society, vol. 16,1 (Autumn 1990) 128 - 151·Reprinted with pennission.]

HMNJou",aJ1