Home

 Research
 > Statement
 > Projects
 > Publications

 CV

 Personal

 Contact Me

Ivention as Preparation for Learning
The Help Seeing Support Environment
Off-Task Behavior
Operator Application Reification

Invention as Preparation for Learning

While direct instruction has been shown to be an effective instructional mean, students often acquire shallow knowledge. Furthermore, there is some evidence that direct instruction leads to rigid knowledge and thus makes students less adaptable. Discovery learning, on the other hand, while being engaging, benefits only a small subset of the students (namely, those who manage to discover).

Recent evidence suggests that there is a win-win solution: invention, prior to instruction and practice, has been shown to lead to more robust learning, compared with instruction and practice alone (Schwartz & Martin, 2004), and to be more effective than unstructured discovery (Klahr & Nigam, 2004)

During the invention process students reason scientifically with data:
1. Students first observe phenomena (how does the data behave?),
2. Then they invent symbolic models (what mathematical method can explain this behavior?),
3. Last, students evaluate and debug their inventions (is the invented method general enough to predict other cases too?).
Following the invention process students receive instruction and practice the canonical solutions. This type of instruction was shown to assist all students. Productive invention does not require successful invention - also students who fail to invent correctly benefited from the invention process.

In this line of research we unpack several aspects of the invention process:
 - Measuring learning: What is the effect of IPL instruction on students' domain knowledge, metacognitive learning, and motivation? Furthermore, do students become better scientists in that they acquire better sense-making skills following the invention process?
- Understanding invention: What are the instructional elements and cognitive processes that are necessary during the IPL process? How does knowledge acquired during IPL differ from knowledge acquired during show-and-practice?
- Scalability: Can the IPL process be facilitated using technology? Furthermore, can Intelligent Tutoring Systems give adequate feedback without reducing students' agency or hurt the exploratory nature of the invention?

Selected Publication: Roll 2009
[top]

The Help Seeing Support Environment

While working with tutoring systems, students are expected to regulate their own learning process. However, often, they demonstrate inadequate metacognitive process in doing so. For example, students often ask for help too frequently or not frequently enough.

In this project we investigate whether cognitive learning principles can be applied in teaching metacognitive skills. More specifically, we focus on improving students' help-seeking skills. We built the Help Seeking Support Environment, which includes the following components:

   1. Direct help seeking instruction using a video, given by the teacher
   2. A Self-Assessment Tutor, to help students evaluate their own need for help
   3. The Help Tutor - a domain-independent agent that can be added as an adjunct to a cognitive tutor. Rather than making help-seeking decisions for the students, the Help Tutor teaches better help-seeking skills by tracing students actions on a (meta)cognitive help-seeking model and giving students appropriate feedback.

In a series of in vivo experiments, the Help Tutor accurately detected help-seeking errors that were associated with poorer learning and with poorer declarative and procedural knowledge components of help seeking. The main findings were that students made fewer help-seeking errors while working with the Help Tutor and acquired better help seeking declarative knowledge components. Furthermore, students demonstrated better help-seeking behavior on subsequent tutor-units, once support was removed. However, we did not find evidence that this led to an improvement in learning at the domain level or to better help-seeking behavior in a paper-and-pencil environment. We pose a number of hypotheses in an attempt to explain these results. We question the current focus of metacognitive tutoring, and suggest ways to reexamine the role of help facilities and of metacognitive tutoring within Intelligent Tutoring Systems.

Selected publications: Roll 2007a, Roll 2007b, Roll 2005.
[top]

Off-Task Behavior

Two threads of work are included under this title:
  • "Gaming the system" is defined as making progress within the curriculum without thinking through the material. Most commonly, within Cognitive Tutors, this is done by guessing repeatedly, or asking fur quick successive hint requests until the answer is given by the system. For more details, see Ryan Baker's description.
  • Learning Goals within Intelligent Tutoring Systems - In this project we extract students' learning goals from their interaction with the tutor (using an ACT-R model).  This information can be used to adapt the instruction to students' learning goals.

Operator Application Reification

In this line of research we investigate what the sources of difficulty for composite problems are. Even though problems involving several skills are harder than those involving only one skill at a time, our results suggest that it is the lack of sufficient proficiency on the single-skill problems that makes the multiple-skills problems more challenging.
[top]