Forward Dendritic Spikes - Advancing Artificial Intelligence Through Biological Process Applications - page 46

out of the reach of synaptic currents integrating
in a passive cable (Mel, 1999).
But things get here even much more exciting. In
the late years we have learnt that intrinsic currents
are not stable components of dendritic membranes.
In fact, they can be up or down regulated by in-
hibitory synaptic currents, internal metabolites,
extracellular effectors, and genetic mechanisms
(Colbert and Johnston, 1998; Tsubokawa and Ross,
1996). The time windows used by such diverse
modulators range from submillisecond to weeks or
longer. As a consequence, the internal processing
of synaptic inputs by intrinsic currents may, and
in fact does change with numerous supracellular
physiological variables, such as the metabolic
state, previous activity, and age. Although we are
only at the beginning of a major field of research,
we can already say that the deterministic view of
dendritic integration can be left to rest.
These breakthrough observations bear im-
portant theoretical implications. We are used
to think of neurons as having one type of input
(synaptic currents) and one type of output (axonal
spikes), while inhibitory synaptic currents are
sometimes classified as negative messages and
others as modulators of excitability (global or
local). At present, the overall view can be quite
different. Excitatory inputs are not the only events
carrying the informative load during synaptic
integration. Lets think for a while on the com-
putational significance of the fact that intrinsic
currents can be modulated. One might well think
of their modulating factors as additional input
to the computing dendrite complementary to
synaptic currents. Thus, dendrites would have
several types of inputs acting on different sites
and with different time scales. Even if we abide to
the simplest of the schemes, we must now admit
that synaptic integration is, at least, a game for
three partners, excitatory, inhibitory and, the new
boy in town, the intrinsic currents. To realize of
their relevance in the behavior of neurons let's pay
attention to this well known fact: when synaptic
inhibition is blocked in a brain nucleus, even si-

lent neurons begin to fire regularly at very high
frequency. This continuous output is meaningless
for target cells (epileptiform behavior), a disaster
that results whenever excitatory and inhibitory
inputs loss balance beyond the physiological
ranges. Such meaningless and intense firing is
brought about by the continuous uncontrolled
activation of intrinsic dendritic currents unleashed
by disinhibition. Since dendritic channels are all
throughout the dendritic surface, we may envis-
age this intrinsic machinery as the actual target
of synaptic inputs, which would operate as local
switches turning on and off specific subsets of a
complex distributed program in the dendritic tree,
as if parallel subroutines in a large multipurpose
software package.

DENDRITIC SPIkES

Are They Real Entities Or Lab
Products?

In the previous section, slow subthreshold intrinsic
currents activated or recruited by synaptic inputs
are credited as important elements defining some
computational properties of dendrites. Yet, a
more intriguing function is their participation in
the initiation of local action potentials or spikes
in the dendrites of many neurons, so called
because they generally fail to propagate to the
axon. Dendritic spikes were initially considered
an anomalous behavior of dendrites caused by
improper observation or illness. It recently be-
come clear that local spikes can be recorded under
physiological conditions, but their significance
still remains controversial. The conflict arises,
at least in part, from the dramatic implications
their presence would have on the very core of the
synaptic integration doctrine. Indeed, local spikes
break up with the classic idea of a single trigger
zone at the axon initial segment.
The current thinking lingers between accep-
tance and skepticism. As recordings are more