Embodied Learning: Moving From Multicamera Mo-Cap to the Kinect

• In Event: Assessing Learning in Embodied Mixed-Reality Environments

Abstract

SMALLab
Mixed-reality embodied learning platforms are coming of age. Several studies have demonstrated increased learning when students were in embodied, mixed-reality environments compared to learning in regular instruction environments where teacher and content were held constant. The MR environment we know best is called SMALLab (Situated Multimedia Arts Learning Lab); it is an educational platform that engages the major modalities (including visual, auditory, and kinesthetic). SMALLab uses twelve infrared motion tracking cameras to send information to a computer about where a student is in a floor-projected environment. Students step into the active space and use a “wand” (a trackable object) that allows the physical body to function like a 3D cursor in the interactive space. With turn-taking, entire classrooms with 30 students are able to physically experience a learning scenario within a typical class period.

The entire ecology of the classroom changes when a traditional lesson becomes a collaborative MR lesson. We believe that the primary drivers for the learning changes we have documented are 1) embodiment, 2) collaboration, and 3) the fact that the experience is still novel. In addition, the gains also may be mediated by language use and gameplay. We have evidence from teacher and student discourse in a chemistry experiment with three High School classes (Johnson-Glenberg, Birchfield, Koziupa, & Tolentino, under review) using a waitlist design. See Table 1.

Our data support the hypothesis that language is affected by the environment. When in SMALLab, 100% of student discourse was on-topic and related to the content to be learned. When students were in small groups working in regular instruction (a project-based, wet lab lesson) only 66% of the content-per-discourse-turn was task-related. Language in mixed-reality environments appears to be more on-topic and learning directed, this may be related to the collaboration built into the design. In addition, students know they will shortly be in front of the entire class performing. Because all students will eventually be “center stage”, they are extremely motivated to get it right. Students receive immediate visual and sonic feedback regarding the veracity of their choices and errors can be quickly corrected. There is something about the affordances of a mediated, co-located collaborative process when combined with gameplay that alters language-use in a classroom. The on-topic language, in turn, affects learning gains. The learning, in turn, affects the flow of the gameplay and these variables continuously interact to create a powerful learning loop.

The Future
Statistical tests using traditional methods are most powerful when a large N is used, however, the current, hardware-heavy motion capture environment is stationary and only one physical classroom in the school can be used (making it almost impossible to do hierarchical linear modeling). Large N studies have been elusive and this is one reason why we are so excited by the possibilities of the cost-effective, more mobile Kinect system (a skeletal-tracking input device that can be used with any modern interactive whiteboard). An example of STEM content taught using the Kinect camera will be presented.

Authors

• Mina Catherine Johnson-Glenberg, Arizona State University

• David Birchfield, Arizona State University