The computational power of formal models for
networks of spiking neurons is compared with
that of other neural network models based on
McCulloch Pitts neurons (i.e. threshold gates)
respectively sigmoidal gates. In particular it
is shown that networks of spiking neurons are
computationally more powerful than these other
neural network models. A concrete biologically
relevant function is exhibited which can be
computed by a single spiking neuron (for bio-
logically reasonable values of its parameters),
but which requires hundreds of hidden units on
a sigmoidal neural net. With regard to the com-
putation of boolean functions we compare the
computational power of spiking neurons with
that of threshold gates, and show the the for-
mer model is more powerful. This appears to be
of some interest, because one commonly views a
threshold gate as the most powerful computa-
tional unit for boolean computations that can
be abstracted from biological neural nets.
This article does not assume prior know-
ledge about spiking neurons, and it contains
an extensive list of references to the current-
ly available literature on computations in net-
works of spiking neurons and relevant results
from neurobiology.