Biomedical Engineering Reference
In-Depth Information
One critical fact is that with maturation, networks acquire functional inhibi-
tory connections, which stops the episodic activity. Mature networks become
segregated into subserving different functions, with a great variety of patterns of
activity. Can the concept of recurrent excitatory network still be used to under-
stand mature networks? In the following we review some evidence from ex-
perimental and modeling studies suggesting that the "immature" mechanisms of
burst generation could be conserved in more mature networks.
First, episodic bursts of activity can be generated in "mature" networks that
are disinhibited by pharmacological block of their inhibitory connections. This
was shown in in vitro spinal (3,10) and hippocampal (45) networks. Therefore,
the immature mechanism is still potentially present in mature networks and can
be unmasked. Furthermore, episodic activity can be generated through the same
mechanism in some networks with functional inhibitory connections if inhibi-
tion is not too strong or if the networks are rendered more excitable
(20,21,33,38-41). Thus, bursting activity could be evoked in mature networks
through neuromodulators that would decrease synaptic inhibition and/or raise
cellular excitability.
Although demonstrating that a mechanism of activity can be uncovered
does not mean that this mechanism is in fact used during the normal function of
a network, there are several examples suggesting such possibility. A model of
the spinal circuit for swimming in the lamprey is based on two excitatory sub-
networks that generate bursts using a cellular adaptation mechanism. These two
units are connected by mutual inhibitory connections, ensuring that the pattern
of rhythmic bursts is in alternation between left and right sides (18). Therefore,
it is possible that the rhythmic locomotor activity is simply a faster version of
the spontaneous activity, with inhibition in the mature network simply allowing
the coordination between left and right sides, as well as between flexors and
extensors in higher vertebrates. Inhibition would also ensure that the locomotor
network is not always "on," but only activated when necessary. Coupled rhythm-
generating circuits control many functions like locomotion, respiration, and
chewing (9), and it is therefore important to understand how these circuits gen-
erate oscillatory activity.
Another example suggesting that the immature mechanism may play a role
in mature networks comes from studies in cortex of anesthetized cat. Timofeev
and colleagues (39) showed that isolating a small slab of cortex led to episodic
bursts of activity (about 5 bursts per minute), with a mechanism similar to the
one presented herein. However, when they recorded from a larger network, they
observed the 1H-z oscillation that is observed during sleep. This led to the sug-
gestion that the cortical sleep (<1 Hz) oscillations and the episodic bursts in
small slabs could be generated through the same type of mechanism.
Finally, an application of this type of activity regarding neural computation
was presented by Loebel and Tsodyks (23). This processing has for its basis the
short "population spikes" generated by networks of mostly excitatory neurons
