Signal Processing Seminar

Title:  Anatomical Cerebrocerebellar Model of Purkinje Cell Simple Spiking 
Based on a Long Loop Cerebrocerebellar Dynamical Control Model

Speaker: Dr. Kazutaka Takahashi

Date:  April 30, 2008
Time:  4:00 - 5:00 pm
Where:  2269 Beckman Institute

Abstract:  

During a horizontal planar manual tracking task, primate cerebellar Purkinje cells 
(PC's) exhibit a wealth of simple spikes (SS) firing activity patterns. However, a 
functional relationship between cellular firing and arm control and its mathematically 
detailed expression had not been established. The nonlinear Recurrent Integrator
Cerebellar Simple Spike (RICSS) model is proposed based upon a recurrent 
integrator proportional-integral-derivative (RIPID) model of cerebrocerebellar 
control. Its structure is based on parsimonious assumption regarding cerebellar system's 
neuroanatomical architecture and its role in long-loop feedback processing. RICSS 
proposed that SS are derived from tracking error signals arriving from cerebral cortex 
modulated by state information carried by parallel fibers in cerebellum.  The error 
signal is projected to a group of cells in a cortical minicolumn to produce broadly 
cosine-tuned signal whose background is modulated by either intended or target speed. 
It is conjectured that one group of parallel fibers (PF), selector PF along with inhibitory 
interneurons carry threshold linear combination of sensed state information modulates 
another group of PF, signal PF that carry the processed error signals from cerebral 
cortex in a multiplicative way. With the threshold linear combination of kinematic states, 
selector PF recruits PC that are particularly tuned for a specific state space. Thus, PC 
SS activity should be related to both feedforward and sensed kinematic information. The 
results reveal the importance of multiplicative dependencies on kinematic signals in PC 
SS data. The particular feature of the RICSS model supports the possibility that 
cerebrocerebellar action occurs as described by a RIPID mechanism that is gain scheduled  
according to behavioral task and sensed limb state.

BIO: 

Dr. Takahashi received his BE degree in Aerospace engineering and Mechanics from 
University of Minnesota, Twin Cities in 1997 and Ph.D in Control and Estimation from 
MIT in 2007. He is currently a postdoctoral scientist at the Department of Organismal 
Biology and Anatomy, University of Chicago. His research interests include gain scheduling 
control, delay system, neural computation and modeling, spatiotemporal analysis of neural 
data, and neurophysiology and neuroanatomy of motor related cerebral cortical areas and 
cerebellum.