Forward Dendritic Spikes - Advancing Artificial Intelligence Through Biological Process Applications - page 48

Available data have been obtained from recordings
in the soma and the thick main apical dendrite
(i.e., the neuron sites where recordings are more
accessible). The erroneous assumption is that these
two recording sites mirror the electrical events in
two other regions, the axon and smaller dendrites,
respectively, which are both hard to access. In
our opinion, this is a dangerous oversimplifica-
tion feeding an artificial dispute. The four com-
partments (axon, soma, main dendrite, and thin
dendrites) each have their own electrical identity.
Let's analyze more carefully this important issue.
While cell firing in the soma reflects faithfully
the axonal firing, it does not provide information
as to the decision zone (axon vs. dendrites). For
example, there is experimental evidence that a
spike coming from the main dendrite jumps to
the axon and only then travels back to invade the
soma, an effect caused by the large surface of the
soma that requires much higher capacitive load
than the slender axon connected to it (López-
Aguado et al., 2002). Thus, the soma always fire
after the axon, regardless of the axonal or dendritic
origin of the spike. Because of this possibility, a
simultaneous recording of the soma (as reflecting
the axon activity) and main dendrite is required.
Yet, there is also evidence that a dendritic spike
may fail to propagate continuously through the
main dendrite and still be able to fire the axon,
so called pseudo-saltatory propagation. In this
case, the output decision has also been made
in dendrites. Time ago, some researchers sug-
gested that specialized dendritic loci with a high
concentration of channels ( hot spots ), possibly in
dendritic bifurcations, may promote this type of
spike propagation (Spencer and Kandel, 1961),
although geometrical factors can also account
for it (Ibarz et al., 2006).
Further on, the main point too often neglected
is that recordings made in the thick primary
dendrites do not mirror the electrical events in
the secondary and tertiary dendritic branches,
where the majority of inputs arrive and interact
with local intrinsic currents. Instead, the behav-

ior of thick dendrites is closely matching that in
the soma, as explained long time ago by pioneer
electrophysiologists (Eccles, Lorente de Nó). They
already noted that the soma and its connected
main dendrites with large diameter work as one,
since the electrical distance is very small. This
forgotten lesson take us to the core of the problem.
When the main dendrite is very long, as in cortical
pyramids, the function changes somewhat and it
behaves as a collector of multiple independent sig-
nals originated in parallel dendritic domains (the
lateral dendritic branches collecting the synaptic
inputs). The questions: Are each of these dendritic
branches capable of generating spikes? If so, do
these spikes enter the main dendrite and propagate
all the way to the soma/axon? Unfortunately, little
is known on how action potentials are initiated in
thin dendritic branches, how are they integrated
locally, and which is the impact on cell output.
There are some indications that local dendritic
spikes outnumber the somatic spikes, and some
are phase locked with them, suggesting that a few
may reach the axon (Kamondi et al., 1998). Curi-
ously, when in vitro preparations are treated as to
reproduce more accurately well known properties
of their in vivo mates, the initiation and forward
propagation of distal dendritic spikes is regularly
found (Williams, 2004). Based on these observa-
tions, the answers to the above questions gain
special relevance as we may be on the wake of
a totally new mechanism of synaptic integration
based on the interplay of local spikes originated
in dendritic substations. The main handicap for
the experimental study is that active dendrites
have a strong non-linear behavior that makes them
highly sensitive to factors hard to control during
experiments. We already mentioned that up- or
down regulation of dendritic excitability depends
on intrinsic factors as well as on the influence of
local networks. Furthermore, many of these fac-
tors are likely to change during ongoing activity,
or under experimental conditions, due to tissue
manipulation and recording or simply through
the choice of preparation.