INC Website NEWS, Fall 2005


New Research Projects

INC principal investigators are initiating several new research projects this year.

Marni Bartlett, Gwen Littlewort-Ford, and Javier Movellan (Machine Perception Lab) received funding from the Naval Research Laboratory to develop systems for automatic Facial Action Coding System (FACS) from video for Homeland Security purposes.  Paul Ekman and Mark Frank in careful laboratory studies have suggested that facial behavior contains reliable information about concealed emotion and deceit.  The system under development at UCSD relies on modern machine learning approaches with state-of-the art performance.  In this four-year project, the UCSD team will develop and evaluate a comprehensive system for fully automating the facial action coding system ($5,137,730.00).
Marni Bartlett and Javier Movellan have also been awarded a collaborative research project with Rutgers University from the National Science Foundation to further develop the Automated Facial Expression Measurement Toolbox and Database ($129,747).

Arnaud Delorme and Scott Makeig (Swartz Center for Computation Neuroscience) received funding from the Mind and Life Institute to investigate neural correlates of meta-consciousness events as experienced during Zen meditation practice ($10,000)
Darpa awarded Scott Makeig a three-year contract to test the feasibility of using EEG scalp recordings of brain activity, together with other psychophysiological and eye gaze measures to improve overall throughput and accuracy of high-throughput scanning of imagery data by intelligence analysts.  This project is in collaboration with Laurent Itti (Computer Science, USC), Pierre Baldi (Computer Science and Biological Chemistry, UCI), and C. Koch (Biology, and Engineering and Applied Science, Calif. Institute of Technology). ($948,194)
Scott Makeig has also received an award from Geo-Centers to purchase equipment to record brain activity using EEG and to characterize four types of EEG dynamic patterns during training task performance: 1) patterns related to the state of arousal and/or alertness of the subject, 2) patterns directly related to the execution of behavioral response selection and execution, 3) patterns related to the level of expertise of the subject developed on the task through the training, and 4) patterns produced by non-brain artifacts include eye movements and muscle potentials.  The goal of the study will be to characterize specific spatiotemporal dynamics that mark progression from novice to expert in a rifle shooting tasks which involves rapid target selection and accurate response ($95,493.00).

Javier Movellan, Marni Bartlett, and Te-Won Lee renewed their Sony-UC Digital Discovery project which explores the idea of humanoid robots that teach children skills and assesses their development while interacting with the children in an affective and human-like manner.  The robot, RUBI, is being developed in close collaboration with personnel at UCSD’s Early Childhood Education Center ($243,207.00).


Howard Poizner transferred his National Institutes of Mental Health (NIMH) grant from Rutgers University to UCSD. Understanding the range of dysfunctions in Parkinson’s Disease (PD) and the degrees to which they are reversible by pharmacological or electrophysiological treatments can increase our understanding of PD therapies and help illuminate critical functions of the basal ganglia in the control of movement.  His experiments explore sensorimotor adaptation and learning in PD ($213,598.00).

In a project funded by the UCSD Kavli Institute for Brain and Mind, Howard Poizner, Scott Makeig, and Terry Sejnowski explore brain dynamics and motor control. The ultimate goal of this project is to better understand the neural control of multi-joint voluntary movement. In the first stage of this research, we plan to use independent component analysis (ICA) and dynamic visualization tools to model the oscillatory field dynamics associated with subjects reaching for targets located in 3D space. To do this, we will record, synchronize and correlate the detailed kinematics of subjects’ arm movements with high density recordings of their EEG in three dimensions. These studies will result in the decomposition of EEG activities into multiple, coordinated regional changes in relation to key aspects of movement.  Understanding the patterns of such movement-related EEG dynamics in normal controls will provide a basis to approach the deficiencies of motor control in Parkinson’s disease and other motor disorders ($30,000).

Terry Sejnowski and Dennis Garlick received support from the Kavli Institute for Brain and mind to explore the neurobiology of abstraction.  Psychological research has observed that abstraction is critical to human intelligence. This suggests that a key goal for brain research should be to identify how the brain achieves abstraction. However, little work has been done so far examining the neurobiology of abstraction. This has likely been at least partly due to the difficulty in getting animals to show abstract performance, as well as the perceived incompatibility between findings in brain research and findings about abstraction from human intelligence research. An approach that can overcome these problems is to examine fluid intelligence tasks. Fluid intelligence is a construct identified by psychometric intelligence research, and measures the capacity of humans to identify and apply abstract relationships. While animals are not able to demonstrate the high levels of abstraction shown by human adults on fluid intelligence tasks, they are able to perform successfully at tasks that are known through psychometric intelligence research to be early indicators of later fluid intelligence. This suggests that the process responsible for abstraction in humans is also likely to be present in a primitive form in these other animals. Developing an animal model of this process will allow an investigation of its neurobiological determinants. This has the potential to lead to a new paradigm that will involve electrophysiology, histology, pharmacology, brain anatomy, genetics, environmental manipulations, and computer simulations ($5,850).