two lies the integrate-and-fire model (Abbott, 1999), which is very common in the
literature because of its ease of implementation while still retaining some biophysically
meaningful parameters.
Researchers knew, however, that neurons did not behave as isopotential units and
therefore incorporated cell morphology into their models. Initially, single dendrites
were considered, with Wilfrid Rail being the first to bring a rigorous mathematical
description of cable theory to neuroscience (Rall, 1959). The basic form of the cable
equation is the same as the heat equation, and hence it is possible to derive analytical
solutions in the absence of active ionic mechanisms, as shown by (Rall, 1959) and by
Wilson (Wilson, 1999). But since such ionic mechanisms gave rise to the nonlinear
spiking behavior that was known to be the method of communication between cells,
modeling pursuits shifted from analytical to numerical methods. Rail pioneered this
change when he introduced compartmental modeling to the neuroscience community
(Rall, 1964).
Multi-compartment modeling soon became standard, allowing both the intricate
dendritic structure and the active membrane properties to be incorporated into the
models. This spatial extension permitted the inclusion of more detailed biophysics,
such as mechanisms describing intracellular processes like calcium exchange, and also
allowed observation of neuronal functions such as back-propagating action potentials.
Such detail has been facilitated by software suites, such as NEURON (Hines and
Carnevale, 2001), which give neuroscientists standardized interfaces through which