|
Neurobiological circuits are rarely well behaved:
- Their activity is noisy, posing a range of questions about how neural computation can remain robust in the face of noise, whether noise can in fact be computationally useful and how neural circuitry shapes this noise.
- Subsets of neurons are known to synchronously oscillate, at a variety of frequencies, depending on the sensory input and behavioral state of the animal.
- The circuit architecture itself is highly recurrent and highly plastic at both short and long timescales.
Although a century of empirical work has produced a wealth of data characterizing these dynamics, a coherent theoretical understanding remains elusive. Purely analytical analysis has proven difficult, as these highly non-linear, non-equilibrium dynamics resist simplification. Modern computational science opens up avenues for analysis using simulation and high-throughput data analysis. I am interested in developing tools to facilitate this.
Using these tools, and in close collaboration with cellular and systems neurobiologists, I would like to build a repertoire of dynamical primitives which the brain is known to use -- a neurobiological "programming language". We can begin to analyze the expressive power of the set of systems which can be built using this constrained language. It is my intuition that by directly employing biophysically-based primitives, exotic though they may be, rapid progress will be made toward a more coherent understanding of cognition.
T. Achler, C. Omar, E. Amir, Shedding the weights: more with less. 2008 International Joint Conference on Neural Networks (IJCNN).
C. Omar, T. Bretl, T. Coleman, Policies for neural prosthetic control: initial experiments with a text interface. 2008 American Control Conference.
C. Omar, M. Johnson, T. Bretl, T. Coleman, Querying the user properly for high-performance brain-machine interfaces: recursive estimation, control and feedback information theoretic perspectives. 2008 ICASSP Invited Session on Brain-Computer Interaction.
|
