Abstract
The model consists of 400 Golgi compartments arranged in a bee-hive-like array and endlessly connected together by parallel fibers. Each compartment consists of one Golgi cell and the surrounding group of granule cells inhibited thereby.
The theory of linear systems was applied to variations from the uniform equilibrium state with the result that any input- or excitation-pattern may be expressed as a superposition of the characteristic patterns, each of which, according to the corresponding characteristic value, has itsown time constant and amplification degree.
The maximum characteristic value λM is positive and is accompanied with striped patterns due to lateral inhibition. The minimum characteristic value is negative and is accompanied with uni-form patterns. Since every characteristic value is proportional to the “inter-layer loop gain” γ, spacial difference in inhibition level as well as AGC-effect for average excitation is intensified as γ increases. The more smoothly the connection weight diminishes with distance, the smaller is λM, and the less dominant is the striping trend.
Results of computer simulation were on the above trends except the conditional occurence of synchronous Golgi firing. This synchronism brought about the oscillating uniform pattern and save way to rather stable and uneven excitation patterns like stripes or mosaics when random noise was independently added to Golgi cells.
Since physiological data seem to negate oscillations, the actual cerebellar cortex is likely to operate in the lateral inhibition mode and the resultant uneven inhibition may decide the contribu-tion of each Golgi compartment to the mossy fiber-parallel fiber pattern transformation. Under such situations, climbingfibersmighttakepart in the information processing of mossy fiber input more or less modifying the inhibition pattern.
