Over a limited range of scale, the dendritic branching of many neurons
is fractal. We propose a growth model, which accounts for the basic p
henomena of neural growth, and reproduces such fractal structures. In
the model, fibrillar proteins diffusing into the cell interact at the
boundary with chemoattractant molecules coming from outside to shape t
he cell. In a parameter range where growth is limited by the diffusion
of chemoattractant molecules, diffuse fractal dendritic trees are obt
ained; analogous to DLA aggregates. For time scales in which growth is
limited by the rate of motion of fibrillar proteins, fractal dentriti
c trees form, but growth is localized to certain surface regions on th
e shape, which resemble experimental growth cones.