CS7001 Projects with Janet Kolodner
Sept. 1, 1999

Four projects combining the areas of cognitive science, artificial intelligence, and learning sciences and technology
(see specific projects for specifics)


Note: All of these projects are big. I don't expect you to do the definitive work for any one of them. I expect you to make a good start. They will serve as an excellent introduction to the kind of research we do in my lab. Anybody who enjoys and does well working on one of these projects is invited to talk to me about joining our project as a research assistant.


1. Promoting transfer the case-based reasoning way (cognitive science, learning sciences and technology)

Transfer is a term educational psychologists use to describe the ability to take something learned in one situation and apply it in another. Near transfer (from a situation to one that is highly similar) is much easier than far transfer (to a situation that is less similar), and there are all different gradations of far transfer. It turns out that it is quite difficult to help learners learn in a way that promotes transfer of what they are learning to new and different situations.

Case-based reasoning means reasoning based on previous experience. It is a computational methodology formulated for building computer systems that could reason intelligently. And it was derived based on inspiration taken from people. We all build on our own previous experiences as we reason. Experience with building a variety of case-based reasoning systems and making their retrieval systems effective provides us with hints about how to promote remembering the right old experiences at the right times - a prerequisite for successful transfer.

In this project, you'll learn about what case-based reasoning has to say about promoting transfer, and you'll take a look at the trajectory of our work in designing software to help with that (JavaCAP and StoryBoard Author), and you'll take this work the next step, doing one of the following:

1. design better software for promoting transfer and justify its design
2. critique StoryBoard Author and design and perhaps implement the changes that would make for a better promoter of transfer
3. design an investigation that would tell us how well StoryBoard Author is working as a promoter of transfer and that would provide us with feedback we need to be able to take it its next steps.


2. Sharing the load of scaffolding between people and the computer; designing an evaluation plan (learning sciences and technology)

Scaffolding is the help one gives a learner to allow him or her to perform at a level beyond what they could do themselves while at the same time providing the kind of guidance that will help the learner learn what it takes to perform at that level without aid. The best scaffolding is aimed at what a particular individual needs. Parents are generally quite good at providing for their children the help their children need to perform better and learn, and they are quite good, often, at providing that help at exactly the right level and at the right times.

It's far more difficult for a teacher of 30 students to provide the individualized help each student needs at the time the student needs it. Still, the idea of scaffolding seems a good one and worthwhile focusing on as we try to figure out how to help learners succeed better.

Several projects are working on designing software-realized scaffolding, in which individuals working on the computer receive the help they need on line. There are a variety of problems with that approach - it is hard for a computer to ascertain exactly what an individual student needs, and it's hard to provide the help in a suggestive way that students will respond to. Others are working on helping teachers provide scaffolding to a whole class or group of students at one time. The problem here, of course, is that not all students need the same help. Still another way to provide scaffolding is to have students who are ahead of others provide it for their peers.

What we actually need, many of us think, is a set of guidelines for distributing the responsibility of scaffolding among all the players - computer, teacher, peers, paper-and-pencil materials. This is what we are trying to do with the SMILE software our Learning-by-Design project is designing. The software prompts students as they write down their ideas and results so they'll do a better job of their writing; it prompts them as they discuss with each other so they'll have some idea about how they might be useful to each other as collaborators; it prompts for other things as well. But since the computer can't evaluate what students are writing, it can't prompt for particular content. That, we suggest, is the role of the teacher and fellow students, and sometimes they need to do it on-line, sometimes as a discussion activity.

We have a way of doing it that we've built into our tools and into the instructions we give the teachers who are using them. But we don't know if we've done the best job. What I want you to do in this project is to read the few conference papers we've written about this project and a set of papers about evaluating educational efforts (papers by Walker and by Brown) and design an evaluation plan that would allow us to find out what we need to know to take the software its next steps (formative evaluation).


3. Indexing a case library of design discussions (intelligent systems, learning sciences and technology, even some hci)

Students using our SMILE software to learn science from design activities have a variety of design discussions as they are engaging in their design challenges. Some discussions are about experiments they've run and the results they've obtained; some are about their design ideas; others are about their design implementations. They present to each other, as well, stories about what they've learned, and the descriptions of cases from the real world.

When a set of students using the system is small, helping students see what others have written is easy - we simply list the set of discussions, and students choose the ones they want to see. If each has an informative title, it's easy for students to choose what they want to read.

However, the real value of using collaborative software for design discussions is allowing those discussions across classes, and even across schools. Once we start doing that, the numbers of groups using the system gets large, as does the number of design discussions. We need a way of indexing those discussions so that students going into the system with some issue in mind can quickly find the design discussion that will help them.

The challenge here is not only in developing an indexing scheme (or set of indexing schemes) that will anticipate the issues that come up in discussions, but in developing one that conforms to the way students think about design challenges and issues, that they can use easily to index the design discussions they begin on-line, and that can be augmented easily while retaining its usefulness by the students using the system.

We have available chapters and papers about indexing (the case-based reasoning way), several sets of design discussions, and the piece of software that helps students manage their discussions. That's the starting point.


4. Promoting habits of mind that promote creativity (cognitive science, learning sciences and technology).

One of my students, Marin Simina, has just about completed a thesis in which he investigated some of the cognition supporting the kinds of creative reasoning done by inventors, scientists, and engineers. He provides clues as to the kinds of control mechanisms that allow for high flexibility (and stubbornness) in reasoning and about the kinds of processing that promote preparation for and recognition of opportunities as they arise.

The big question for us is what it would take to promote those kinds of reasoning habits. We have the beginning of an answer. We need to have students learn these habits in the context of being creative. Our definition of creativity is pretty down to earth. We aren't looking for students to be creating something new and different from what anyone else has done before. Rather, we want them to attempt to achieve complex challenges in thoughtful ways - going beyond the obvious.

Given this partial answer, we think the best place to promote these kinds of habits is in our Learning-by-Design classes, where students are learning science in the context of trying to achieve complex design challenges - e.g., designing the propulsion system for a miniature car to learn about Newton's laws, designing a model of a way of maintaining a beach site (in a stream table) to learn about coastal change, waves, and erosion. And we think it a worthy idea to begin by looking at the software we've created for promoting thoughtfulness, planning, and reflection while working on challenges - SMILE - and to move forward in answering the big question from there.

There are two alternative ways of approaching this challenge:

o Learn enough about creativity (Marin's thesis) and about Learning by Design to begin to make suggestions about how to enhance a classroom environment - lots of cognitive science here.

o Learn enough about creativity, Learning by Design, and SMILE to propose augmentations and/or changes to SMILE that would better promote creative habits of mind.


5. Promoting productive case-based reasoning (intelligent systems, cognitive science, learning sciences and technology)

I'm running out of steam. If you're interested, talk to me about this.