Biomedical Engineering Reference
In-Depth Information
value. This can happen because in the absence of activity s keeps increasing
towards its asymptotic value, 1. As soon as n # s reaches its control value, epi-
sodes of activity reoccur. The intervals between episodes are somewhat longer
than in control, because the exponential increase of s during the intervals is
slower since s is now closer to its asymptotic value. Therefore, the activity can
recover after a moderate decrease of connectivity because the system compen-
sates by decreasing the level of depression (increasing s ).
This predicts that the unblocked synapses see their availability or efficacy
increase relative to their control level. To verify this prediction, we have stimu-
lated a pathway (between adjacent ventral roots) that does not contain glutama-
tergic synapses. Indeed, we have shown that the strength of the response was
increased after blockade of glutamatergic connections and subsequent recovery
of the activity (37). The developing spinal circuits are therefore able to ap-
proximately maintain their level of activity following the blockade of some of
their connections. This is very important since the temporal pattern of activity
may be important in the development of network and cellular properties (15,34).
During development, some cells and connections may be lost, while other syn-
apses may see their efficacy increased. Through the very mechanism that regu-
lates its patterned activity (activity-dependent depression), the developing spinal
cord is able to compensate for these changes and therefore maintain its activity
level within a certain operating range.
5.
DISCUSSION AND FUTURE WORK
We have presented an idealized model of spontaneous activity in develop-
ing neural networks. Despite its mean-field approach, the model captures the
emergent nature of the phenomenon. Although we implicitly assumed that a
small proportion of cells were active, none have pacemaker capabilities, so the
ensemble interactions are crucial for generating the episodic rhythm, not just for
synchronizing cellular oscillators.
The model presented in this chapter was developed to understand the
spontaneous activity in the developing spinal cord. However it is general enough
to apply to spontaneous activity in other developing circuits. Indeed, other
modeling and experimental studies have suggested that similar mechanisms can
explain the spontaneous episodic activity in developing retinal (4,14,24)
and cortical networks (27). These mechanisms involve fast positive feed-
back through excitatory connections together with a slow activity-dependent
depression of network excitability. They may therefore be common to many
developing networks. This gives us a framework to study developing and
excitatory networks. Can this framework be helpful to the study of mature
neural networks?
Search WWH ::




Custom Search