98 Computer

E N T E R T A I N M E N T C O M P U T I N G

A lthough computers can repre-sent a medium for children’ssocial and intellectual devel-opment, some researchersbelieve that using computers

before age seven subtracts from impor-tant developmental tasks and othertypes of learning (J.M. Healy, Failureto Connect: How Computers AffectOur Children’s Minds—for Better orWorse, Simon & Schuster, 1998).Those opposed to computers believethat computer-based activities are lesseffective in developing understandingand skills than are artifacts that youngchildren can handle.

These anxieties extend to technolo-gies such as smart toys. Some research-ers (D.E. Levin and B. Rosenquest,“The Increased Role of Electronic Toysin the Lives of Infants and Toddlers:Should We Be Concerned?” Con-temporary Issues in Early Childhood,vol. 2, 2001, pp. 242-247), for exam-ple, claim that electronic toys producelimited and repetitive interactions thatmight inhibit the healthy play anddevelopment of young children. Theycontend that playing with dolls, trucks,blocks, and similar toys encourageschildren to be the “creators and con-trollers of their play” and helps “par-ents play in imaginative, give-and-takeways with their infants and toddlers.”

CACHETOur recently completed research

project, Computers and Children’sElectronic Toys (www.ioe.stir.ac.uk/cachet/), investigated how children usesmart toys. Cachet combines recentinterest in mobile learning, tangibleinterfaces, and the home use of tech-nologies. This research aimed mainlyto explore interactivity and interfacesin the context of smart toys that chil-dren could use alone or in conjunctionwith a computer.

ToysWe based these toys on Arthur and

his sister D.W., two aardvark charactersfrom the Marc Brown stories and car-toon series (http://pbskids.org/arthur/)familiar to more than 75 percent of thechildren in our study. The toys have aplush finish, resemble traditional softtoys, and—unlike some smart dolls—their sensors and batteries do not makethem heavy and unwieldy. Arthur andD.W. stand 60 cm tall and have avocabulary of about 4,000 words,motors to provide movement, and elec-tronic chips to recognize inputs. Because

the toys cannot respond intelligently tospoken input, they depend on gesturalinteraction. If a child squeezes a toy’shand or wristwatch, the toy will askquestions. If a child squeezes the toy’stoe, it will suggest a game.

Games include estimating a time—5,10, 15, or 20 seconds—by squeezingthe toy’s hand when the time is up, say-ing the alphabet backwards and for-wards, and tongue twisters. In additionto using them by themselves, childrenalso can use the toys in conjunctionwith specially encoded CD-ROMs thatfeature language and number games.

Playing with the toy and the software

simultaneously requires an accessory: aradio transmitter that looks like amodem and connects to the computer’sgame port. A PC pack add-on increasesthe toy’s vocabulary to 10,000 words,letting the toy “talk” to children, com-ment on their interaction with the soft-ware, and offer advice and encourage-ment. Children can elicit help and infor-mation from the toy by squeezing itsear. In this mode, the child interactswith both the computer and the toy,while the toy interacts with the com-puter and mediates the child’s actions.

If two children play with the toy, theinteraction possibilities multiply. If thetoy is not present, a clickable onscreenicon of Arthur or D.W. provides helpand information. If children have dif-ficulty with a game, the toy or iconreminds them of this help. We focusedon these toys because they deliver thesame help content through differentmechanisms.

Conducting the studyWe developed a multidisciplinary

Interactivity,Interfaces, andSmart ToysLydia Plowman, University of StirlingRosemary Luckin, University of Sussex

Despite some technicalhurdles, using smart toys and software for educational play showspromise.

approach to provide multiple response-and-behavior indices appropriate to thesocial and cultural context of using thetoy while also allowing for investigationof detailed interactions. Given that chil-dren can play with Arthur and D.W. bythemselves or in conjunction with thecompatible CD-ROM, we looked at toyuse alone, software use alone, and usingthe two together. Using a common coreof data collection methods across allsites, we looked at toy and computeruse in children’s homes, after-schoolclubs, and classrooms.

Twelve children aged five to six par-ticipated in the home studies and werevisited three times over approximatelytwo weeks. Half the children receivedthe toy first and were given the CD-ROM during the midway visit, whilewe reversed this order for the otherhalf. In all cases, the children kept bothitems for the second week and couldthus play with the toy and PC together.

Classroom use was more controlled,with detailed, dual-source video analy-sis of 32 children aged four to five.Twenty-two children in the four after-school clubs participated in the study.All children received verbal instruc-tions on using the software and vieweda demonstration of how to access thehelp facility. In the classrooms and atthe after-school clubs, the researchersremained in the background butprompted activity when childrenbecame stuck.

Analyzing the dataTo capture the multiple interactions

experienced by pairs of children usingboth the toy and software and eitherone or the other, we used the follow-ing main categories when transcribingthe videos: action and gesture, expres-sion and gaze, and dialogue. We fur-ther divided these categories intobetween children; between toy andchildren; between toy, onscreen soft-ware characters, and children; andbetween children and researcher. Weused these structured transcripts tocode categories of interaction andexplore a range of questions about

children’s help-seeking preferences andsocial behaviors.

We compiled all data relating to theindividual children in the homes andindividuals or pairs in the after-schoolclubs to produce separate case studies.These studies contain personal details,field notes, complete video sessiontranscripts, interviews, the diaries, testresults, and the rating scale. This infor-mation let us analyze data for individ-uals and across children while buildinga detailed picture that includes the con-text of use, individual differences, pat-terns of play, and toy and software use.

RESULTSAlthough the study focused on one

particular type of toy, our approachhas proven valuable in laying thegroundwork for future research.

Toy characteristicsInterview questions revealed a con-

siderable diversity of views about theextent to which children attributedhuman characteristics to the toys.Some knew batteries powered the toy,but younger children tended to thinkthe toy had feelings and could thinkand talk by itself.

Invited to suggest ways in which thetoys could be improved, several chil-dren mentioned that they would like itto be able to walk, perhaps becausethis would make the toy more lifelike.Although some children seemed toimbue the toy with sentient qualities,parents’ comments suggest that thisdid not translate to a greater degree ofdialogue or other forms of interaction.Most children viewed the toy as justthat and some preferred to play withit switched off, taking the toy to thedinner table or making a bed for it nextto the child’s own.

At the first visit, parents tended tofind the toys scary or feared they wouldstifle imaginative play. Yet by the endof the loan period, half the parentsclaimed the toys provided some edu-cational value and said they would buyone at an affordable price. Althoughparents found the ways in which thetoy and software interacted impressive,children seemed to take it for granted.

In homes, the time spent playingwith the toy or software decreased asthe novelty diminished—from about45 minutes per day to less than 15 min-utes, with interest in the toy wearingoff faster than interest in the software.Those who received the toy afterreceiving the software played with itless frequently, if at all, partly becausethe help features were not generallyneeded—the software had alreadybeen explored. However, in thissequence we also found that childrenviewed the toy mostly as an adjunct tothe software and rarely played with itaway from the PC.

Children sometimes placed the toyon a table, next to the computer screen,even though the radio transmitter hasa four-meter range. Although this madethe toy seem more like a peripheralthan an interaction partner, placing thetoy on the child’s lap could interferewith mouse manipulation. Paired chil-dren avoided this problem because onewould typically control the mouse andthe other the toy.

Interacting with toys and softwareAlthough the toys can verbally inter-

act with the children at a basic level,the spoken interaction could detractfrom the possibility of extended child-toy interaction and role-playing.Initially intriguing, the toy’s vocabu-lary presents only an illusion of reci-procity and seems too limited to implyreal personality. Most children foundthe toy’s talking monotonous or irri-tating and eventually switched it off.

We found no evidence that these toysmake either a beneficial or detrimentaldifference to the children’s ability toengage in child-led imaginative play.

February 2004 99

Half the parents claimedthe toys provided some educational value and

said they would buy one at an affordable price.

100 Computer

E n t e r t a i n m e n t C o m p u t i n g

To understand how children use thehelp available to them, we combinedthe descriptive results across contextsand with detailed activity analysis fromthe school studies. Initially, childrenwere more likely to seek help fromhuman companions. Although theyoften failed to notice the unsolicitedclues the toy or onscreen icon was giv-

ing, when prompted by their humancompanion, they became competent atusing the toy to elicit hints and encour-agement. The children were discerningusers and recognized the help content’squestionable value.

The toy’s presence showed a statis-tically significant increase in incidencesof help being successfully imple-mented, with children actively seekinghelp and adults actively engaged inprompting or assisting the children intheir software use. However, the toyand no-toy conditions made little dif-ference in how often children refusedor ignored the offered help.

These findings suggest that systemdevelopers interested in software scaf-folding might reap benefits from con-sidering tangible interfaces instead ofscreen-based ones. The onscreen iconand the toy provided the same feed-back content, with only the method forinvoking and delivering this feedbackdiffering.

W hile all the children in ourstudy enjoyed using the soft-ware, interest in the toy

appeared to be age-related, with par-ents reporting much more interest fromyounger children. Their poor feedbackmakes these toys unimpressive inter-action partners. Nevertheless, the tech-nology has potential, and generalizing

from findings relating to the specifictoys in the study suggests several areasfor future development.

Using existing work on softwarescaffolding could improve the type andmode of feedback by linking it to thechildren’s performance. Because theyoungsters were more likely to interactwith each other or with the researcherwhen the toy was present, tangibleinterfaces offer promise for improvinginteraction between peers.

Children or their families failed todiscover some of the toy’s functionality,such as the alarm clock. This shows thetradeoff between a toy being complexenough to maintain interest and simpleenough for use by very young children.

In preschool settings and the earlyyears of primary school, the computeris usually used as a free play activitywithout the benefit of adult mediation.There may be value in developing thistechnology for such circumstances,but this would require close analysisof the contingent help adults can pro-vide. Likewise, the plausibility issuemust be overcome. Finally, girls’attraction to the toy could provide anavenue for redressing an imbalance inthe greater appeal of technology toboys than girls. �

Lydia Plowman is Reader in Educa-tion and Director of PostgraduateResearch in the Institute of Education,University of Stirling. Contact her atlydia.plowman@stir.ac.uk.

Rosemary Luckin is director of theHuman Centred Technology ResearchGroup at Sussex and the InteractiveDigital Educational ApplicationsResearch Lab. Contact her at rosel@sussex.ac.uk.

The children enjoyed the tactile natureof the toys more than the toy’s inter-activity, which challenges the view thattechnological toys are psychologicallydamaging.

Detailed video analysis revealed thatyoung children can make the connec-tion between two different interfacesand coordinate the experience receivedthrough their convergence. Our evi-dence suggests that children as youngas age four are not disconcerted whenfaced with feedback and interactionpossibilities from different artifacts.

Many children required assistancefrom the researcher or a peer to elicithelp from the toy or onscreen icon.Some children also ignored the helpthat Arthur or D.W. provided. Evenwhen they took notice of the helpprompts, however, they did not neces-sarily interpret them correctly.

Help featuresWe analyzed the toy’s role in sup-

porting the child’s learning in terms ofscaffolding: how a more knowledge-able partner can assist the cognitivedevelopment of a less able one andgradually foster the development ofsuccessful independent task perfor-mance. Considerable research hasexplored this concept in relation tosoftware, but with emphasis on desk-top computers. We extended thisresearch by examining how childrenrequested and used assistance from thetoy, the accompanying software, orother people.

Editor: Michael Macedonia, Georgia Tech Research Institute, Atlanta; macedonia@computer.org

Girls’ attraction to the toy could help redress animbalance in the greaterappeal of technology to

boys than girls.

Renew your

IEEE

Computer Society

membership

today!

www.ieee.org/renewal