NEURON is designed for empirically-based neural modeling of biological neurons and networks, and is particularly well-suited for use by experimentalists. NEURON also handles models of artificial spiking neurons, which are simulated using an extremely efficient discrete event method. Network models may contain any combination of biophysical and artificial spiking model cells. As of September 3, 2010, there were more than 1000 scientific publications that reported work performed with NEURON (see the list of publications at http://www.neuron.yale.edu/neuron/static/bib/usednrn.html).
NEURON runs on all leading computational platforms from workstations and PCs (UNIX/Linux/OS X and MSWindows) to supercomputers. In a parallel processing environment it can be used to simulate distributed models of neurons and networks, i.e. simulations in which different parts of the model cell or network are handled by different processors. Extensive tests using published models of individual conductance-based neurons and networks of such neurons have been run on parallel hardware (e.g. individual PCs with multiple CPUs, workstation clusters, or massively parallel supercomputers). These tests demonstrate speedup that is at least linear with the number of CPUs, until there are so many CPUs that each one is solving fewer than ~100 equations.
NEURON was created by Michael Hines and John Moore at Duke University, and Michael has been in charge of its active maintenance, development, and extension ever since. In recent years, NEURON has adopted an open source development mechanism to help meet the evolving needs of neuroscientists. NEURON is distributed at http://www.neuron.yale.edu and http://neuron.duke.edu.