Wang, XD (reprint author), Chinese Acad Sci, Shanghai Inst Biol Sci, Grad Sch, Inst Neurosci, Shanghai 200031, Peoples R China,email@example.com
Numerous learning rules have been devised to carry out computational tasks in various neural network models. However, the rules for determining how a neuron integrates thousands of synaptic inputs on the dendritic arbors of a realistic neuronal model are still largely unknown. In this study, we investigated the properties of integration of excitatory and inhibitory postsynaptic potentials in a reconstructed pyramidal neuron in the CA1 region of the hippocampus. We found that the integration followed a nonlinear subtraction rule (the Cross-Shunting Rule, or CS rule). Furthermore, the shunting effect is dependent on the spatial location of inhibitory synapses, but not that of excitatory synapses. The shunting effect of inhibitory inputs was also found to promote the synchronization of neuronal firing when the CS rule was applied to a small scale neural network.