'Smart' toys helping to make smart kids: a different type of interface

Marketing strategies for the toys from Microsoft had included claiming educational value for them, with each of them being described on the box as an 'interactive learning partner'. The study did not measure learning gain but some measures for the standard proxies for learning, such as engagement, motivation and time on task, are available from the diaries and interviews. Based on these, it is unlikely that any short-term learning gain would be sustained, as interest in the toy diminished over a relatively short period.

While initially intriguing, the toy's vocabulary presents only an illusion of reciprocity and appears to be too limited for the necessary suspension of disbelief for a child to imply personality. Most children found that the talking became monotonous or irritating, preferring to switch it off. The home study children were older than those in other contexts and this may account for parents' comments on how quickly children's interest in the toy had waned or that the toy had proved more popular with a younger sibling (although all were well within Microsoft's suggested age range).

Detailed video analysis revealed that young children are able to make the connection between two different interfaces and co-ordinate the experience they receive through their convergence, including those who gained low scores on the WPPSI-R tests. Our evidence suggests that children as young as four are not disconcerted when faced with feedback and interaction possibilities from different artefacts. Many children did, however, require assistance from the researcher or a peer in order to elicit help from the toy or on-screen icon and there were examples of children ignoring the help provided by Arthur or DW. The children in the study were able to understand the mechanics of the toy interface and all could engage at an operational level (eg mouse control and the relationship between the mouse and the on-screen cursor). However, whether with the toy on its own or in combination with the software, not all children engaged in activities at the conceptual level. Unlike adults (Fogg & Nass, 1997), they are not taken in by the constantly positive and flattering feedback and even if they take notice of the help prompts, they do not necessarily interpret them correctly.

The results from the teacher's assessments (PPBS) confirm the traditional view that girls spend less time using the class PC and less time trying to work out how things work However, the girls showed the toy more affection and responded more appropriately to the toy's help prompts. From the classroom data we can see that the girls spent more time using the software when the toy was present; conversely, the boys spent more time when the toy was not present.

The technology also allows peers simultaneous access to interactions with the content and activities in a way that the standard desktop computer does not. Both children can interact with the software, one via the toy and the other via the keyboard or mouse, whereas children operating the PC by themselves are often so absorbed in the software that they pay little or no attention to the toy. However, when pairs use the toy and software together, successful interaction depends on both children working together as a team. The mouse clicks override the toy's sensors and if this is not recognised, the child controlling the toy can become frustrated and bored.

The role of the toy in supporting the child's learning was analysed in terms of scaffolding, a term coined by Wood et al. (1976) to account for how a more knowledgeable partner can assist the cognitive development of a less able one and gradually foster the development of successful independent task performance. This study extended the existing research by examining the ways in which children requested and used assistance from the toy, the accompanying software, or other people (Luckin et al., under review and 2003, forthcoming).

From this it will be possible to construct an understanding of children's use of the help available to them. The children in this study were more likely to seek help initially from human companions: a parent, the researcher or a fellow peer. In fact, they often failed to notice or process the unsolicited clues being given by the toy or the onscreen icon but if prompted by their human companion they became competent at using the toy to elicit hints and encouragement. The children were discerning users and recognised that the usefulness of the content in the available help was questionable.

Ineffective or irritating feedback from the toy, particularly wholesale praise and flattery, was not welcomed and on some occasions became a source of irritation and a distraction, particularly for children with high WPPSI-R scores.

One of the games which appeared to provide the impetus for children to engage with the help available was a searching game in which the tasks asked of the child were often discrete and offered a clear goal. When the toy was present, t-tests showed that there were statistically significantly more incidences of children successfully implementing the help they were offered. There was almost no difference between the toy and the no-toy condition in the number of times children refused or ignored the help offered.

This increase in social interactions around technology contradicts the popular belief that technology leads to reduced socialisation. It suggests that system developers interested in creating a Zone of Proximal Development between less and more able learners through software scaffolding might reap benefits from considering tangible, as opposed to screen-based, interfaces. The content of the feedback available through the onscreen icon and that available through the toy is the same; the method by which this feedback is invoked and delivered is different.

The future of interface contingency may lie in a system with alternative interfaces from which an assistant selects the most appropriate for a required intervention. Recognising interface contingency would represent another case of seeing if human teaching strategies are amenable to computer-based assistance.

All children seemed to enjoy the software, regardless of ability or age, but the low level of interest in the toy appeared to be age-related: parents reported much more interest in the toy from younger siblings. These toys are not impressive as interaction partners because the quality of the feedback is poor, often leading to frustration and irritation.

Nevertheless, the technology has potential and generalising from findings relating to the specific toys in the study suggests the following areas for future development.

• The type and mode of feedback could be improved by using existing work on software scaffolding to link it to the children's performance.
• As the children in the study were more likely to interact with each other or with the researcher when the toy was present, tangible interfaces are a promising way of improving interaction between collaborating peers around technology.
• Some of the toy's functionality was not discovered by children or their families (eg using it as an alarm clock). There is a trade-off between a toy being complex enough to maintain interest but simple enough for very young children to use.
• In pre-school settings and the early years of primary school the computer is usually seen as a free play activity (Plowman & Stephen, 2003b) and children use it without the benefit of adult mediation. There may be value in developing this technology for such circumstances but this would require close analysis of the contingent help that adults can provide and the plausibility issue would need to be overcome.
• The attraction of the toy for girls may provide an avenue for redressing an imbalance in the appeal of technology for boys and girls.