Análisis comparativo de los índices generales de sustentabilidad para las

3.2.1 Active Electrodes

Voltage is the potential for current to flow from one place to another, consequently there is no such thing as a voltage at a single point (Luck, 2005). Therefore an ERP waveform reflects the voltage difference over time between an active and a reference

electrode (the reference electrode is discussed in the next section). However, directly measuring the voltage difference between two electrodes would reveal any surplus electrical charges that had built up in the participant and would obscure any neural signals. To solve this problem a differential amplifier is used. A differential amplifier uses three electrodes to record activity: an active electrode placed at the chosen site on the scalp, a reference electrode placed elsewhere on the scalp, and a ground electrode placed at a location on the scalp or body. The voltage from the ground electrode is subtracted from the active and reference electrodes and the amplifier then amplifies the difference between the active and reference electrodes [(active minus ground) minus (reference minus ground)]. Any electrical charges picked up by the ground electrode will be the same for the active minus ground and reference minus ground calculations and will therefore be eliminated by the subtraction.

In practice, one active electrode is rarely used in isolation. Simultaneous recording from a montage of active electrodes covering multiple scalp locations is necessary to quantify distinct ERP components that may be maximal at different scalp sites. The use of multiple recording sites allows ERPs to be differentiated on the basis of their

distribution (topography), and has the additional benefit that eye movement artefacts are also more readily observed (Picton et al., 2000).

The placement of electrodes in the montage is typically based on the International 10-20 system (Jasper, 1958). This system uses features of the skull (the nasion, inion etc.) to position the electrodes on the scalp and assumes that the skull is symmetrical.

However, although this assumption is rarely met, variability in electrode placement due to skull asymmetry does not appear to be large enough to result in alignment with different underlying cortical structures across participants (Binnie et al., 1982; Homan et al., 1987). The 10-20 system accommodates up to 75 electrodes. The larger the

montage the greater the spatial resolution and hence greater accuracy in detecting topographic differences. However, there appears to be little difference in spatial resolution between montages of 64 electrodes and 128 electrodes (Tucker, 1993;

Srinivasan et al., 1998) consequently due to the increased time to apply a montage of 128 electrodes, a montage of 64 electrodes was used in the research reported in this thesis.

3.2.2 Reference Electrode

If the reference electrode were to pick up the brain activity recorded by the active electrode, this activity would be cancelled out when the difference between the electrodes was calculated. Thus, although no site is truly neutral, it is essential to use the most neutral possible reference site. Previous ERP studies of recognition memory have generally used the bony prominences (mastoids) behind each ear. The mastoid references are easy to apply and are not distracting for the participant. To avoid a hemisphere bias, electrodes are placed at both the left and right mastoid and the wires are physically linked to create an average of the two mastoid electrodes as a reference (Miller et al., 1991). In practice, the EEG is often recorded using a left mastoid

reference, allowing the quality of the right mastoid to be observed on-line. The left and right mastoids are then algebraically reconstructed off-line to create a linked mastoid reference, circumventing two potential problems associated with recording using a linked reference. First, linking the wires creates a zero-resistance electrical bridge between the hemispheres, which distorts the distribution of voltage over the scalp (Katznelson, 1981). Second, if the reference electrodes had different impedances (electrical resistance), then the linked mastoid reference would move toward the

electrode with the lowest impedance and produce hemispheric bias (Miller et al., 1991).

The position of the reference electrode determines the morphology of the EEG

waveform recorded at each active site. Amplitudes at active electrodes proximal to the reference electrode are attenuated more than amplitudes at more distal active electrodes.

The research in this thesis uses the linked mastoid reference to facilitate comparison with previous episodic memory research.

3.2.3 Analogue-digital (A/D) conversion

The voltage difference between each active electrode and the reference electrode is recorded as an analogue signal therefore, as computers require digital signals, it must be converted into digital form. To do this, the analogue signal is firstly amplified (this is necessary because the voltage detected at the scalp is small) and then passed through high-pass and low-pass filters that remove activity that is not within the range of normal EEG (0.01 – 40 Hz). The high-pass filter passes high frequencies but attenuates low frequencies [e.g. electrogalvanic (skin) signals], whereas the low-pass filter passes low frequencies and reduces the amplitude of high frequencies [e.g. electromyographic (muscle) signals].

The analogue signal is then converted to a digital signal using an analogue-to-digital converter. The converter samples the analogue signal at discrete time points. The sampling period is the amount of time between consecutive samples (e.g. 8 ms) and the sampling rate is the number of samples taken per second (e.g. 125 Hz). The Nyquist Theorem is needed to decide the sampling rate to use. This theorem states that all of the information in an analogue signal can be obtained digitally as long as the sampling rate is at least twice the highest frequency present in the analogue signal. If the signal is sampled at a rate lower than this, information will not only be lost but artifactual low

frequencies will be induced in the digitised data (this is known as aliasing; Picton et al., 1994).