Event-related simulation of neural processing in complex visual scenes

Understanding the neural processing of complex visual scenes is facilitated by tools that allow efficient simulation of large neural systems. We here present an environment for the implementation of large networks of generalized integrate-and-fire neurons which uses an asynchronous event-based algorithm. It allows accurate (up to machine precision) calculation of neuronal spike times. Its efficiency arises from its ability to solve the subthreshold dynamics of the model analytically. The simulator is implemented in JAVA and it can run in parallel on a large number of computers with different architectures without the need for compilation. The neuronal network to be simulated and all parameters are defined in extendible markup language. A model of the primate early visual system is implemented. The use of the tool is illustrated by simulating the processing of both simple and complex visual scenes through retina, thalamus and primary visual cortex.