An MIT Media Lab Research Project

Building a Learning Companion

Go to the Learning Companion Web page at the MIT Media Lab

The current educational system is constructed so as to emphasize the evolution of rule-based reasoners-students that understand ‘what’ is to be done, or ‘how’ a task is done-how to do recursion, diagnostic reasoning, cognitive assessment, etc. is, at best, a minor part of the curriculum. This trend is becoming more pronounced as the Education Reform initiatives of various states mandate the acquisition of factual information and the achievement of abilities to perform certain mechanical skills. These initiatives do not provide for the development of an appropriate measure of model-based reasoning and creative problem solving skills. Our meta-hypothesis is that scientists, mathematicians, engineers, and technologists (SMET) are clearly not rule-based reasoners/thinkers. They ARE model-based reasoners/thinkers. They possess what we refer to as learning intelligence (e.g., the ability to do: recursion, visual modeling, diagnostic reasoning, cognitive assessment, and cognitive appraisal). If we are to spawn SMET learners, educational pedagogy must be revised to infuse model-based reasoning skills in our school systems in as part of the pedagogy and part of the learning journey.

We propose to discover, describe, and evolve the cognitive and affective learning processes required for SMET learning based upon our meta-hypothesis. We then intend to incorporate these research-based findings into a testbed simulation-the learning companion (a software-based interactive application), which will recognize the affective and cognitive state of the learner and respond in a manner. We expect our results to be applicable to computer-based artifacts (e.g., the learning companion, companion-like software applets built into curricular software), and to impact the pedagogical approach of educators.

This study is particularly interesting as it provides a rapid transition from hypothesis generation (SMET learners must be model-based reasoners rather than rule-based beings) to hypothesis testing by capitalizing on the development of a new learning model. For example, with the assistance of an innovative observational tool-the learning companion, we expect to test our hypotheses that: varying the learning environment in response to the transient emotions of the learner will improve learning intelligence, attending to the affective state of a learner and responding in an appropriate/intelligent way will significantly improve the quality of learning, knowing how to spawn the most critical emotions in order to facilitate maximal results can, for example, spawn the teachable moment, and, understanding the processes a learner experiences during an ‘off-goal’ episode (e.g., cognitive assessment, cognitive appraisal, emotions) will enhance SMET learning.