The Poizner Motion Capture Lab and Unsupervised Learning

 

MocapWhen someone enters the Poizner Motion Capture Lab, it might first appear that they have time traveled to a futuristic world. Here, a person can actively explore large-scale, texture-rich, complex virtual environments. While the subject moves freely in this virtual world, the lab combines motion capture and high density EEG to record body and limb movements, as well as eye movement. The goal of this virtual reality lab? To study the neural control of action and learning.

 

Dr. Poizner, TDLC member and Sensory Motor Network leader, explains, "We are studying what we call 'unsupervised learning' - that is, how subjects learn about the world through self-exploration rather than through explicit instruction." He explains the process: "Subjects navigate in these virtual towns while we record their movements and their EEG. We study how these patterns change over the course of learning. In conjunction with several other lines of research, we will be able to build up information about the neurobiology underlying this very important, very natural, very commonplace human form of learning (unsupervised learning). We will be able to understand the brain patterns and their relation to the changes in action and exploration patterns over the course of learning. Eventually we will be going from single cell level to synapse level, to behavioral level, to combined EEG and fMRI imaging level."

 

In order to create this fully immersive VR system, the lab uses high-end devices such as a 24 optoelectronic camera system to monitor 3D movements; a 12-monitor head mounted display to present the virtual environments; a 72-channel EEG system, and a 20 speaker 3D sound systems with a centrally controlled Mocap avatarsound server. This advanced technology allows the lab to record a subject’s body movements and render the subject moving as a computer avatar in real time in the virtual environment. Dr. Poizner explains that these virtual environments are very powerful tools because "we have total control over the environment. We can gain insight into the nature of the representations that subjects form of the world by, for example, altering the world once it has been learned, seeing how the movement pattern has been changed and how the neural pattern has been changed. Thus, we can see how the subject is able to generalize the knowledge to new situations, and we can see how the neural dynamics change in response to those novel inputs."

 


Additional Information

The Poizner Motion Capture Lab projects, funded by the Office of Naval Research, The Temporal Dynamics of Learning Center, NIH, and the Veterans Administration, involve active collaborations with other units on UCSD campus as well as with national and international investigators. For more information about the Poizner Motion Capture Lab, please click here. To view a video: please go to http://inc.ucsd.edu/~poizner/media.html.