2.1.1. Electrogenesis
The general structure of a typical neuron is illustrated in Figure 2.1. The neuron is surrounded by a neuronal membrane containing cytoplasm and the nucleus. The cytoplasm, which is also referred to as the intracellular fluid, consists mainly of water and electrolytes (electrically charged molecules and ions). The membrane works as a barrier between the intracellular and the extracelluar fluids and controls the flow of ions entering and exiting the neuron, which in turn determines the difference in voltage between the inside and the outside of the neuron. This ability is maintained by the protein molecules that the membrane is made from. Some of these molecules are attached to the surface of the membrane whereas others penetrate the membrane and create a bridge between the inside and the outside of the neuron. These bridges are known as ion channels.
Presynaptic cell
Figure 2.1 The basic structure of a neuron.
Features include the dendrites, cell body and axon. Action potentials travel down the axon in the direction indicated by the arrows. Information is exchanged between neurons at the synapse; action potentials cause neurotransmitters to be released from the presynaptic cell and bind to receptors in the postsynaptic cells causing ion channels to open or close. This reaction results in a postsynaptic potential: graded change in potential across the membrane.
When a neuron is resting, the separation of positive and negative charges across the cell membrane sustains an electrical potential of approximately -70 mV (by definition, the outside of the neuron has an electrical potential of 0). The negative resting potential is primarily caused by a higher concentration of potassium ions in intracellular compared to extracellular fluids (due to the large numbers of open potassium channels in the membrane). When a neuron is stimulated, the electrical potential rapidly changes and, if the neuron is depolarised sufficiently, the result is an action potential that propagates to the terminal of the neuron. The action potential works on an all-or-nothing basis; as long as the neuron’s potential reach a certain threshold, the electrical impulse will be initiated to its full intensity. The sudden change in voltage in one area in the axon of the neuron will elicit a similar reaction in a nearby area and in this way the impulse will travel the full length of the neuron in the manner of a chain reaction. It is important to note that the only matter that actually moves along the axon during this progression is the electrical current;
the ions move restrictively in and out of the cell membrane and the surrounding fluids stay in position. When the action potential reaches the terminal of a neuron, chemical neurotransmitters are released at the synapse. The neurotransmitters fit into receptors at the dendrite of the post-synaptic neuron. When the neurotransmitters combine with the receptors this causes ion channels to open or close resulting in a graded change in potential across the membrane, known as a postsynaptic potential. Hence, action potentials reflect transfer of information within a neuron (intracellular potentials), whereas post-synaptic potentials reflect transfer of information between two or more neurons (extracellular potentials).
Although action potentials can be measured using invasive single-unit recordings, they are generally not registered by scalp electrodes (Luck, 2005) because neurons that are aligned in parallel to each other are likely to send action potentials down the axons at the same time. This synchronisation would not be a problem if the action potentials were triggered in perfect synchrony, but this is often not the case. When there is a slight time delay, one neuron will be letting ions out through the membrane when another neuron is letting ions in at the same spatial location. The action potentials then cancel each other out and therefore produce a signal that is too small to be detected from the scalp (Luck, 2005).
Post-synaptic potentials, on the other hand, last longer than action potentials, are typically restricted to the one location (the dendrites) and arise instantaneously.
Post-synaptic potentials are therefore the signals that are picked up by EEG recording electrodes placed on the scalp. It is important to note, however, that for
the signal to be detectable, a relatively large population of neurons must a) fire simultaneously, and b) be arranged in an “open field” geometric configuration. In open field configurations neurons are aligned in a parallel orientation and when the population of neurons fire simultaneously, the electrical fields generated by each neuron will sum together. A great proportion of the cerebral cortex is structured in this way (Coles & Rugg, 1995). However, neurons in some regions of the brain, especially subcortical structures, are arranged with the cell bodies clustered in the centre and dendrites reaching out in all directions. Such an arrangement is known as a “closed field” configuration, and activity from neurons aligned in this manner is very unlikely to be picked up by scalp electrodes (Coles & Rugg, 1995). One critical factor that follows from the selective sensitivity of EEG to particular types of neural activity is that when no difference in ERP activity is present as a function of experimental manipulations, one cannot confidently conclude that no such differences exist because they could simply be invisible at the scalp.
2.1.2. Volume Conduction
ERPs inherently provide less accurate spatial information compared to haemodynamic imaging methods (such as fMRI and PET), because they only measure signals from the surface of the head. EEG activity recorded from the scalp can be the result of a near infinite number of intracerebral sources that cannot easily be identified; a problem which is known as the “inverse problem”. The main reason for the poor spatial resolution is that the inside of the skull acts as a volume-conducting space. The electrical signals are smeared out as they pass through the brain, severely distorting the voltage distribution as it appears on the surface. The
signal recorded from a location on the scalp depends on the position and orientation of the neural generators as well as the resistance and shape of the brain and the skull.
The inverse problem is the reason why ERPs are not an ideal methodology for investigating the various anatomical structures underlying cognition. However the distribution of ERP activity across the scalp still contains some valuable information. For example, in cognitive research, it is sometimes sufficient to determine whether or not the neural processes observed in two experimental conditions are engaged by the same or different neural systems. In the case when two experimental conditions give rise to ERPs of differing topographic distribution, it is reasonable to conclude that different sets of neural generators are engaged across the conditions (or at best, that there is differential engagement of generators).
Unfortunately, since an infinite number of dipoles can give rise to the same pattern of voltage distributions, meaning that when no topographic differences are present it is still possible that different subsets of generators are involved across conditions. It is, however, important to emphasise that the source localization of EEG signals can be estimated based on MRI and head models. Source localization was not attempted in this thesis because the primary focus was kept on the temporal characteristics of memory and metamemory-related ERP activity rather than the anatomical structures involved.