Biomedical Engineering Reference
In-Depth Information
1.
INTRODUCTION
A key problem in neuroscience is to understand the cellular basis of behav-
ior. Vertebrate neural networks have very diverse architectures, and are com-
posed of large numbers of neurons of different types, connected through various
classes of synaptic contacts. In order to discover the principles underlying the
activity of neural networks, it may be helpful to understand their history. There-
fore, a possible strategy is to study developing networks, as they might be sim-
pler to study but nevertheless operate through similar mechanisms as the more
complex, mature networks. One major advantage of this strategy is that many, if
not all, developing neural networks generate a similar type of activity that has
been called "spontaneous activity." Therefore, the conclusions obtained from the
study of one network may be of general application.
In this chapter, we review the characteristics of this spontaneous activity,
focusing on the networks of the developing spinal cord. We then present an ide-
alized model of this activity and discuss two important applications of this
model. Finally, we discuss the generality of this model and its possible applica-
tion to more complex networks.
2.
SPONTANEOUS ACTIVITY IN DEVELOPING NETWORKS
Early in development, neuronal networks of the central nervous system
generate spontaneous activity. It is called spontaneous because it is not provoked
by sensory inputs or inputs from other parts of the nervous system, but is gener-
ated within each of these circuits in isolation. Spontaneous activity has been
well characterized in the developing spinal cord, hippocampus, and retina and
has also been described in other circuits (25). Although they have widely differ-
ent architectures, the features of the spontaneous activity are very similar in all
these networks (25). The most characteristic feature of this activity is its
epi-
sodic
nature: most if not all neurons of the network become active together for
several seconds to a minute; then the network becomes silent for intervals that
can last up to several minutes, as illustrated in Figure 1.
Because spontaneous activity is so widespread in the developing nervous
system and with striking similarities between different circuits, understanding its
mechanisms of generation may provide some general principles of neuronal
network function. Furthermore, there is evidence that spontaneous activity can
drive the refinement of neuronal circuits (17,46), as this activity usually involves
large populations of neurons in a highly correlated fashion (see below) and
therefore may lead to strengthening/weakening of synaptic connections through
Hebbian mechanisms (28) (see (46) for a discussion of spike-timing-dependent
potentiation/depression of synapses related to developing circuits). Finally, the
temporal pattern of activity may also regulate the electrical properties of indi-
