Microworlds, Explanatoids and extending the Islands of Expertise Theory

I’m reading David Williamson Shaffer’s book “How Computer Games Help Children Learn”

In it, he refers to a paper that he and Gina Svarovsky published on in 2006 based on their Digital Zoo project.  I had a hunt for this paper and found a number that they published on this, and an earlier project called Berta’s Tower.

Both projects seem to have been structured play / learning experiences based around the excellent physics / creative toy SodaConstructor.

I’ve just finished reading one paper ( SodaConstructing an understanding of physics: Technology-based engineering activities for middle school students ) based on the Berta’s Tower project.  

Some (unrefined) initial thoughts are below…

In this paper, Shaffer and Svarovsky present their analysys of the learning experiences of 10 children as they performed a series of tasks that aimed to improve their knowledge of gravity and centre of mass physics.

IMHO, the paper achieves a few things.

• it extends the “Islands of Expertise” model of learning developed by Crowley and Jacobs (2002) to include the concept of exploratoids and by doing so, provides a link to the “Design, Build, Test” cycle that is used in industry.
• it shows the usefulness of exploratoids as a construct to understand the experience of students who are engaged with a simulatory ‘microworld’
• it highlights what the authors perceive as the key factors in delivering a successful exploratoid based learning experience
• it provides some empirical evidence of knowledge gain through play (although it’s not particularly ‘clean’ given the lack of a control group)

The Experiment

In the study, Shaffer and Svarovsky ran two workshops with 6 students in each.  They conducted pre and post interviews for quantatative analysis and filmed session for transcripts and qualatative analysis.

In the workshops, they had pupils try to perform a task (such as “build a multi story tower”).
Half way through the task, a discussion was held with the students and their design team (it’s not clear who the design team was) to share progress.
At the end of each task, a ‘best design’ was shown to all other pupils and the items arising from the tasks were discussed.

Results

Shaffer and Svarovsky analysed the data by coding responses using “ten emergent categories”.  It seems a bit weird to me to code the responses emergently like this - it almost feels as though there was no original purpose to the tasks pupils were given.  If there was an agenda - if there was a goal in performing these tasks, would you not attempt to code the responses to show improvement (or a lack thereof) in this regard?

That aside, the quantitative results were positive, with players expressing significant improvement in their understanding of centre of mass as well as using much more scientific language to do so than in the pre-interviews.

Extending the Islands of Expertise Model

The quantitative results are not the most interesting aspect of this paper however.

As mentioned above, Shaffer and Svarovski propose to extend Crawley and Jacobs’ Islands of Expertise model by including exploratoids in addition to the original explanatoids.

Islands of Expertise

According to Crawley and Jacobs (via this paper - I haven’t read the original), Islands of Expertise are topics that children happen to become interested in and develop a “relatively deep and rich knowledge”.  From what I infer from Shaffer and Svarovsky, Crawley and Jacobs use Islands of Expertise as a way to understand and formalise the natural phenomenom of children taking an interest in and becoming ‘experts’ on a particular topic.

Explanatoids

Crawley and Jacobs explain these “islands” via what they term explanatoids.  An explanatoid is a short fragment of explanatory conversation that a parent shares with a child.  Each explanatoid is individually unremarkable but they stimulate interest and, over time, can foster a “motivating and powerful connection between interest and understanding”.

So children have an inkling of an interest in something and that interest is encouraged and deepened through informal conversations with their parents.

Exploratoids

This is where Shaffer and Svarovsky extend Crawley and Jacobs’ model.  They define an exploratoid as “short fragments of exploratory action between a student and microworld that over time accumulate to build interest and understanding.”

So an exploratoid is an action that the player expresses that results in a small modification of the player’s mental model of the domain and which, over time can develop a deep understanding of the domain in the player.  This reminds me of the view portrayed in Raph Koster’s excellant (and fun) book “A Theory of Fun for Games Design“.

It seems reasonable to me to extend the idea of explanatoids to exploratoids.  I agree with the authors that they provide a useful construct to expain the learning taking place when a player engages with a game ( / simulation / microworld - call it what you will…).  I have tried to express the concept many times when explaining the e-Bug games to others but I have lacked appropriate nomenclature.  So, yay for jargon

Design Tips

The paper has a transcript of two boy’s interactions with SodaConstructor during the workshop.  Shaffer and Svarovsky use this to explain the attributes of the process (and the tool) that allow this kind of learning to take place.

Autoexpressivity

The microworld (game or simulation) embeds domain rules in the entities and relationships it expresses.  When players interact with the game by acting upon these entities, the microworld responds appropriately.  In other words, if the domain is ‘Newtonian Physics’, then an dropped object should fall.  As such, “the behavior of the tool reflects the extent to which the student can represent the underlying domain principles with the grammar of the tool.”

And therefore as players revise their actions within the game to achieve a goal, “they also test and revise their understanding of the embedded domain.”

Expressivity

Shaffer and Svarovsky also champion the freedom of players to choose how to handle a task.  The game world shouldn’t be completely prescriptive and should afford the player open-ended choices on how to handle the task at hand.  As they say:

“When microworlds are used in open-ended activities, they allow students to develop understanding through expressive projects: that is, projects that allow students to explore their own individualized design decisions, to create solutions that are inventive, unique, and personalized.  This freedom to explore can be both meaningful and motivating for students, affording them a sense of control and personal investment in their inquiry”

Iteration

By allowing the player to iteratively design, build and test various solutions to a problem, the autoexpressive nature of the simulation delivers the exploratoid.

It is the iteration combined with the player’s expressive actions and the simulations autoexpressive responses that lead to deep knowledge and the “Island of Expertise”.

Yeah.. But.. No.. But..

So this is interesting for me.  In designing the e-Bug games, I have tried to avoid the “chocolate covered broccoli” approach of rubbish game based learning in favour of embedding the learning right into the gameplay.  I agree with Koster’s understanding that playing a game == learning and have tried to design my games so that the game mechanics are true to my learning outcomes.

Shaffer and Svarkovsky come only a little short of saying that you need to have an autoexpressive, expressive, iterative microworld in order to implement games based learning via exploratoids.  This worries me a little!

I have designed the e-Bug games with the same basic idea of learning through exploring the world’s mechanics.

• the e-Bug games are autoexpressive - they attempt to reflect the relationships of the embedded domain
• the e-Bug games are iterative - they allow the player to try something, see that it didn’t work and try again with a subtly different strategy.
• the e-Bug games are not particularly expressive.

I have been much more prescriptive and restrictive in how the player does engages with the world.  My assumption has been that seeing the consequenses of player actions would be sufficient to deliver learning.  There are some options at the player’s disposal, but not so many as to be considered open-ended.

This makes me a little concerned - if it is true that a game seeking to deliver learning through exploratoids must be not only iterative and autoexpressive but also expressive then I fear that I may be in a hot spot!