Information in the brain is carried by the temporal pattern of action potentials
generated by neurones. The patterns of spike discharge are determined by intrinsic
properties of each neurone and the synaptic inputs it receives. Modulation of either of
these parameters changes the output of the neurone and through this the behaviour of
physiology of the organism. Computational models of brain function usually focus on
how patterns of connectivity contribute to information processing, but fail to take the
different intrinsic properties of different neuronal phenotypes into account. Models of
single neurones that take into account all intrinsic mechanisms will be extremely
complex, and hence building large-scale models of neurone networks will be
computationally intense, if not infeasible. In order to develop simple models that still
reflect realistically the intrinsic properties of the neurone we first need to know which
of the many identified mechanisms are the most important for its function.
Conventionally, intrinsic properties are investigated in detail in isolated cells
in vitro. Insights gained thereby are taken to speculate how these mechanisms
contribute to spike patterning or neuronal responses in vivo. However, in vitro
experiments are performed under artificial circumstances. Besides the relative scarcity
of afferent input in dissociated cells, the preparations for the experiments require
interventions that fundamentally disturb cell properties. Thus, it is problematic to
interpret observations made in vivo on the basis of results obtained in vitro.
In this thesis a novel, radically different way of investigation is presented. We
examine recordings of firing activity of oxytocin neurones using statistical methods.
The use of spontaneous, unexceptional activity recorded in vivo avoids all the
interventions and alterations associated with in vitro preparations, also allowing to
take the influence of afferent input into account. The main purpose of our work is to
determine key features involved in the regulation of discharge patterns, and consider
possible explanations in terms of known intrinsic properties. A further objective is to
determine whether there are any consistent, characteristic differences in the firing
pattern of oxytocin neurones recorded under a variety of physiological conditions
(naive, pregnant, lactating, and hyperosmotic stimulation).
We have found that while firing activity appears to be random (except for the
effects of the HAP) on a small time scale (>0.5 s), on a time scale of several seconds
it appears to be much more ordered. Also, we found evidence of a 'balancing'
mechanism, whereby on a short to medium time scale periods of faster activity are
followed by periods of slower activity (and vice versa), thus leading to a rather
homogenous and steady activity overall. Of the known intrinsic mechanisms the AHP
effects the firing activity in a way compatible with the firing characteristics found.
Thus, from the results of the statistical analyses we conclude that the most important
parameters to determine the firing of oxytocin neurones are the post-spike HAP and
the post-train AHP.
In addition, the analysis of the activity recorded under different physiological
conditions reveal that the firing of pregnant and hyperosmotically stimulated
neurones is remarkably similar to the firing in nai've organisms. In contrast, firing
activity during lactation shows subtle differences indicating that the AHP decays
faster in these circumstances, which is in agreement with results obtained in vitro.