The hippocampus resembles the neocortex in containing a majority of excitatory neurones, the pyramidal cells, and granule cells, which use glutamate as their neurotransmitter (E in Fig. 2.3.9.1).^{(1, 2)} Most of the remaining 10-20 per cent of neurones in the hippocampus are inhibitory, and use γ-aminobutyric acid (GABA) as their neurotransmitter (I in Fig. 2.3.9.1). The inhibitory neurones fall into several distinct subtypes according to where their axons go (and hence which cells they inhibit), where their cell bodies are, the shapes of their dendrites, whether they contain more than one transmitter, and whether they contain particular calcium-binding proteins. This chapter will ignore most of the diversity of interneurones.⁽⁵⁾

There are many more excitatory pyramidal cells (E and triangles) than inhibitory interneurones (I and circles). As with most neurones, they receive inputs onto their dendrites and somata (the latter contain the nucleus and are represented by a triangle or circle). The level of simplification is clear from the observation that each pyramidal cell receives tens of thousands of synapses. Axons from other regions, known as afferents, make excitatory synapses (e) with both pyramidal cells and interneurones. Most of the interneurones make inhibitory synapses (i) onto pyramidal cells. The inhibition of the pyramidal cell is called ‘feed-forward’ (f.f.) when the interneurones were excited by afferent axons and ‘feed-back’ (f.b.) when they were excited by pyramidal cells. Interneurones also inhibit each other forming a mutually inhibitory network (m.i.); this network is important in some kinds of physiological network oscillation (see text). Finally, pyramidal cells make excitatory synapses onto each other (r.e.), which can lead to epileptic discharges if not held in check by inhibitory mechanisms.

Each hippocampal (or neocortical) area receives ‘afferent’ synaptic inputs from other areas. Most afferents are excitatory; stimulating them provides a convenient tool to study the operation of the neuronal circuits involved. The responses evoked in hippocampal neurones typically start with an excitatory postsynaptic potential. If the excitatory postsynaptic potential is strong enough, it will result in an action potential triggered at a low-threshold zone near the cell body, probably a short distance down the axon. The excitatory postsynaptic potential is followed by a fast and a slow inhibitory postsynaptic potential. Both the fast inhibitory and excitatory postsynaptic potentials are due to ligand-gated channels where the transmitter receptor is part of the same molecular structure as the ion channel. In the case of inhibition this is the GABAA receptor, which allows chloride ions to pass. In the case of excitation it is a variety of glutamate receptors, which are permeable to sodium, potassium, and in some cases calcium ions, and which are further subdivided into α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)/kainic acid, N-methyl-D-aspartate (NMDA) and other classes. The slow inhibitory postsynaptic potential is due to GABA_B receptors, which are G protein coupled and use second messengers to open separate potassium channels. Many other kinds of G-protein couple receptors exist, and often are involved in modulating neuronal excitability or synaptic function.

Inhibitory neurones can be triggered both by activity in the principal cells (pyramidal or granule), resulting in recurrent or feedback inhibition, and directly by the incoming afferents, resulting in feed-forward inhibition. Experimentally, the synchrony of the stimulation of the afferent input imposes synchrony on the response with the useful consequence that the extracellular currents generated by the activity of individual pyramidal or granule cells can summate (because the cells are located in tight layers) and produce large ‘field potentials’ comprising a population excitatory postsynaptic potential, followed by a population spike.

Field potentials evoked by local stimulation are over in 10 to 20 ms and the slowest intracellular components end within a few 100 ms to 1 s. However, stimulation can have much more prolonged effects. The best known of these is long-term potentiation, in which a brief train of stimuli can result in an increase, lasting hours or days, in the response to a fixed test stimulus. The modest conditioning event and the enduring consequence make long-term potentiation an attractive model of learning and memory, although the evidence that it really is the direct cellular substrate for learning remains circumstantial.⁽⁶⁾ It is perhaps more likely that long-term potentiation provides an artificial experimental tool that depends on cellular and molecular mechanisms that may also be involved in learning and/or other plastic changes in synaptic strength.

Hippocampal neurones are not just arranged as a simple synaptic relay where afferents excite target cells to produce an output depending on the size of the input and the state of inhibition at that time. Instead there exists a complex synaptic network, or local circuit, interlinking neurones of all kinds (Fig. 2.3.9.1 gives a very much simplified illustration of some of the salient features of hippocampal local circuits). This chapter considers two kinds of emergent network activity that arise from this organization: focal epilepsy and gamma rhythms.