The brain ability to perform meaningful signal processing tasks related to perception, pattern recognition, reasoning is normally attributed to large-scale neuronal networks. The main signals involved in the instantaneous neural processing are neural impulses, and the units, which process impulses in a network, are individual neurons. We now put a question: In the context of higher brain functions, like perception, what is a meaningful task a neuron performs with the signals it receives? Another question: Does the inhibition exist for taming neuronal activity only, or it can be endowed with a more intelligent signal processing role? In this article, we propose an abstract concept of signal processing in a generic neuron, which is relevant to the features/events binding well known for large-scale neural circuits. Within this concept, action of inhibition obtains its natural signal processing meaning.
Hodgkin and Huxley Equations
If denotes the displacement of the transmembrane potential of the excitable membrane from its resting state, then its time course is defined by the transmembrane currents as follows:
(1) where
is the number of different ionic currents considered,
is the capacity of the membrane unit surface and
,
denote ionic currents through that surface. In the Hodgkin and Huxley (H-H) model, (Hodgkin & Huxley, 1952), three currents were considered, namely, the potassium, sodium and leakage current. These currents depend on the
by the following way:
(2) where
,
are time-independent. The so called gating parameters
depend on
in accordance to the following equations:
(3)Here parameters, ,depend on in a nonlinear manner, see (Hodgkin & Huxley, 1952) for the exact expressions. The system (1)–(3) has resting state with . The temporal dynamics is usually introduced into (1)–(3) through a choice of proper initial conditions with a nonzero value. This corresponds to experimental manipulation known as the voltage clamp method. After the voltage clamp is released, the temporal dynamics of can be observed either experimentally, or by solving (1)–(3) numerically.
The remarkable feature of the H-H set of equations is that if the initially clamped value of corresponds to depolarization and is high enough, then the dynamics itself builds up further depolarization up to a definite value, , and then returns to its resting state. This transient process is known for real neurons as the action potential, or spike, and it constitutes the essence of the neural impulse, when propagates along the membrane of a neural fiber, (Hodgkin, 1971). Both for real neurons, and for the set of Equations (1)–(3) neither the time course of the action potential, nor its peak value does depend on the initially clamped value of . Moreover, the time course of the action potential obtained by solving (1)–(3) is in perfect correspondence with that observed experimentally for the giant nerve fiber of squid (see Hodgkin & Huxley, 1952, Figs 13, 14).
The ideas of H-H equations received further development in several directions. First, additional ionic currents found in other neurons and the dynamical properties of corresponding ionic channels are added to the (2) and (3) (Huguenard & McCormick, 1992). Second, spatially distributed (compartmental) equations are considered in order to fit with morphology of real neurons (De Schutter & Bower, 1994). Third, for simplification of mathematical analysis, a reduced sets of equations were offered, which has lower dimension than (1)–(3), and still is suitable for generating spikes (FitzHugh, 1961).