microvolt (a millionth of a volt) and therefore require considerable amplifica-
tion (in the order of several tenth of thousand times) before they can be
converted into a digital stream that can be displayed on a screen and stored
on a computer disk for further analysis. EEG amplitude can be recorded every
millisecond (or even with a higher sampling rate if required) from 64 or 128
electrodes simultaneously. The temporal resolution of EEG is therefore in the
same range as the speed of neural function.
To go from EEG to event-related potentials (ER Ps), it is required to know
and record the time at which ‘events’ (e.g. stimuli or responses from partici-
pants) have occurred. This is achieved by a signaling system involving trig-
gers, that is digital markers that registers the onset of an event and becomes
part of the EEG recording (see Figure 9.2 for examples of triggers sent from
the stimulus computer to the acquisition computer). To obtain ERPs from
EEG, several processing stages are required, which generally involve (a) filter-
ing (to reduce the contamination of the EEG by environmental and mechani-
cal noise not originating in the brain); (b) ocular / heart artifact reduction or
correction; (c) visual inspection for suppression of remaining artifacts; (d)
epoching (i.e. cutting the continuous EEG recording into pieces or epochs
starting a little before event onset and generally finishing before the next
significant event); (e) baseline correction (which allows differential pre-event
potentials at different electrode sites to be eliminated); (f) averaging of EEG
epochs for each experimental condition of interest within (individual ERPs)
and across (grand-average ERPs) participants; (g) re-referencing ERPs to a
reference of choice, depending on the electrode(s) used as a reference during
recording; (h) computing mean amplitudes and peak latencies from various
parts of individual participant ERPs; and (i) conducting statistical analyses
to test for significant differences in amplitude or latency between experi-
mental conditions.
The rationale behind ERPs is that spontaneous activity produced by the
brain over the scalp is not synchronized to events such as stimulus presenta-
tion or participant response unless it is related to it. By averaging EEG epochs
recorded in response to a stimulus (e.g. a word or a picture), for instance,
brain activity that is unsynchronized (i.e. irrelevant for stimulus processing)
will progressively disappear as the number of epochs increases because this
activity will be randomly positive or negative, depending on trials, and over-
all average to zero. By contrast, brain activity that is related to the presenta-
tion of a given stimulus (we talk about time- or phase-locking) will
progressively average in measurable variations as the number of epochs
grows. Such variations are not arbitrary and in fact they are highly reproduc-
ible from one experiment to the next in a given individual. Even though ERP
amplitudes are essentially idiosyncratic to each participant, the same overall
morphology of ERPs can be identified in most individuals, leading to a
nomenclature of the main variations or peaks based on polarity (negative/
positive) and order of appearance or rounded latency.
Language Selection During Speech Production in Bilingual Speakers 207