We present a novel approach to modeling neural behavior using a "qualitative reasoning" algorithm. The Qualitative Reasoning Neuron (QRN) is capable of qualitatively reproducing single neural networks without loss of critical details. QRN simulations of a single Purkinje cell (approx 1600 compartments) show significant speedup over a recent GENESIS model. A large scale model of the cerebellar cortex (256 neurons, approx300,000 compartments) is used to simulate a saccadic eye movement task. The model reproduces in vivo Purkinje cell bursting patterns during saccades. We simulate rapid and gradual adaptation paradigms and show that error correction is possible when climbing fiber input is periodic and contains no error signal.
展开▼