As discussed in the previous chapter, the Pavlovian type of learning involves only the presynaptic neuron. In the Hebbian type of learning, the synaptic modification is induced by the participation of both presynaptic and postsynaptic neurons.

Recalling that learning is the ability to associate two events that happen almost at the same time. In the hippocampus, events are represented by a population of neurons, each may be either excited or in the resting state. A particular event is represented by a particular set of neurons in the excited state. For instance, if the number of neurons involved in the representation is n, then mathematically an event can be denoted by a vector with the dimension n, 

X = [x₁, x₂, x₃, .....x_n]

where x_i (i = 1 - n) is either "0" (resting) or "1" (excited). 

The hippocampus contains the most complex neural network. Each neuron is connected to thousands of other neurons. For simplicity, we assume that only five neurons are involved in the representation of events. The connection of these neurons are shown in Figure 7.1. The lines are drawn from the nerve terminals (presynaptic) in the upper row to the dendrites (postsynaptic) in the lower row. For example, in Figure 7.1, the terminals of the first neuron are connected to the dendrites of the second and fourth neurons. The terminals of the second neuron are connected to the dendrites of the first and third neurons. These lines may also be considered as the synapses between neurons. The "dark cloud" is represented by the excitation of the second and fourth neurons and the "rain" is represented by the excitation of the first, fourth and fifth neurons.

Figure 7.1. Illustration of the Hebbian type associative learning.

In the Hebbian type of learning, the synaptic modification may be induced only when both presynaptic and postsynaptic neurons are excited. In Figure 7.1, these synapses are represented by red lines. The left red line connects the second and first neurons; the right red line connects the fourth and fifth neurons. When the dark cloud and rain happen almost at the same time, these neurons are excited and their synapses are modified so that nerve impulses can be more easily transmitted from the presynaptic neuron to the postsynaptic neuron. Suppose before learning the nerve impulse is unlikely to transmit from the second neuron to the first neuron. After the pairing between dark cloud and rain, this synaptic transmission is greatly enhanced. Next time, the dark cloud alone is likely to cause the excitation of the first neuron, thereby increasing the probability to recall rain.

How could physiological system implement Hebbian type of learning? The answer lies in the NMDA channel, which is a subtype of glutamate receptor channels. For most synaptic channels, activation (opening) requires only the binding of neurotransmitters. However, activation of NMDA channels requires two events: binding of glutamate (a neurotransmitter) and relief of Mg²⁺ block. NMDA channels are located at the postsynaptic membrane. When the membrane potential is at rest, the NMDA channels are blocked by Mg²⁺ ions. If the membrane potential is depolarized due to excitation of the postsynaptic neuron, the outward depolarizing field may repel Mg²⁺ out of the channel pore. On the other hand, binding of glutamate may open the gate of NMDA channels (the gating mechanisms of most ion channels are not known). In the normal physiological process, glutamate is released from the presynaptic terminal when the presynaptic neuron is excited. Relief of Mg²⁺ block is due to excitation of the postsynaptic neuron. Therefore, excitation of both presynaptic and postsynaptic neurons is necessary and sufficient to open NMDA channels.

Another important feature of the NMDA channel is that it conducts mainly the Ca²⁺ ion which may activate various enzymes for synaptic modification. The enhancement of synaptic transmission is called long-term potentiation (LTP), which involves two parts: induction and maintenance. The induction refers to the process which opens NMDA channels for the entry of Ca²⁺ ions into the postsynaptic neuron. The subsequent synaptic modification by Ca²⁺ ions is referred to as the maintenance of LTP. We have just explained the mechanism for the induction of LTP. The maintenance of LTP is accomplished by insertion of ionotropic receptors (e.g., AMPA receptors) into postsynaptic membranes.

AMPA receptors are a subtype of glutamate receptors. They may form ion channels in the postsynaptic membrane to conduct small cations. Higher density will generate greater ionic influx when glutamate molecules are released from the presynaptic terminal, thereby resulting in more membrane depolarization, which in turn facilitates transmission of the nerve impulse from a presynaptic neuron to a postsynaptic neuron. It has been shown that Ca²⁺ can activate calcium/calmodulin-dependent protein kinase II (CaMKII) and drive AMPA receptors into postsynaptic membranes (reference). This memory trace, however, cannot last very long, because AMPA receptors are constantly cycled into and out from postsynaptic membranes due to a constitutive pathway and busy activities (review 1; review 2).

While the formation of short-term memory is now well understood, conversion from short-term to long-term memory remains largely unknown. The next few chapters will present a possible mechanism, wherein microtubules play a central role.