Hiroaki GOMI, Munetaka SHIDARA, Aya TAKEMURA, Yuka INOUE, Kenji KAWANO, Mitsuo KAWATO
Using an Inverse Dynamics Representation to Reconstruct Temporal Firing Patterns of
Ptukinje-cells in Monkey Ventral Paraflocculus
Abstract:The complex temporal firing frequencies of Purkinje cells (P-cells) in the monkey cerebellum ventral paraflocculus were reconstructed based on a second order linear inverse dynamics representation consisting of eye acceleration, velocity, position, and bias terms. These firing patterns were observed during ocular following responses (OFR) induced by movements of a large-field visual scene. Many P-cell firing patterns (109/232 from 30 P-cells, 16 stimulus conditions) were well reconstructed from eye movements with reliably estimated coefficients by least square error minimization technique with evaluating the modeling error, the fitness of the reconstructed firing frequency and the reliability of the estimated time-lag. Further, two statistical methods show the necessity and sufficiency of the second order model.
From the coefficient analyses, the two types of P-cells (local linearly and global linearly correlated to the eye movements) were found although the nonlinear relations were observed between the stimulus speed and firing frequency in the most of P-cells. The coefficients of a single P-cell independently estimated in several stimulus conditions were close to each other in the P-cells global linearly correlated to the eye movements, and these temporal firing frequencies in several stimulus conditions were successfully reconstructed by one parameter set. This result indicates the generality of this linear inverse dynamics representation for these P-cells. In the local linearly correlated P-cells, the firing frequency tended to be insensitive as the eye velocity increased.
Moreover, the averaged ratio of the acceleration coefficient to the velocity one is close to that in eye motor neurons, suggesting that the dynamic component of the eye driving signal for ocular following response is generated by the P-cells. This result supports the hypothesis that the cerebellum works as a motor command generator which computes an inverse dynamics model for each motor system.
Key words: ocular following response, cerebellum, Purkinje cell, firing frequency reconstruction, inverse
dynamics representation, least square error method.