1.2 Justification for an Alternate Method for Quantifying Single
Trial Measurements: A Peak-Picking Approach
Peak picking is defined as the identification of the largest positive or
negative peak amplitude in the averaged ERP in a pre-determined time
window (Weeda et al., 2012; Zhang & Luo, 2011). The time window used to
define a given ERP component can be based on time windows used in the
extant literature. If prior studies using similar stimulus paradigms are
unavailable, the morphology depicted in the grand averaged ERP (i.e.,
the average of all averaged ERPs of all participants in the study) can
be used to determine a time window that captures the component of
interest for the peak picking of the averaged ERPs of each individual
participant. The advantages of the peak-picking methodology are that it
allows one to 1) obtain an amplitude and latency measure of the
component of interest at its maximum point, 2) measure differences in
latency between individual participants or the conditions in the study,
and 3) attain a greater sensitivity (i.e., lower measurement error) in
the peak amplitude measurements when compared to averaging voltages in
the window. Additionally, if each major peak of an averaged ERP is
measured, then the obtained amplitude and latency measures define a
generalized morphology of brain processing to the event for each
participant.
Once a peak is identified in the peak-picking approach described above,
there are two ways that the amplitude can be calculated. One approach to
peak picking has been to measure peak amplitude from the baseline
(a.k.a. baseline-to-peak measures) after performing the processing step
of baseline correction. While baseline correction aids in normalizing
the segmented trial data to changes in voltage from a zero-voltage
referent, this procedural step can also make the peak sensitive to
background noise, especially in single trials (Weeda et al., 2012). The
second approach, measuring amplitudes from peak-to-peak may be less
sensitive to baseline noise and, therefore, a better way to measure
amplitudes with single trials (Ribeiro, Paiva, & Castelo-Branco, 2016;
Unsal & Segalowitz, 1995). However, it has yet to be identified how
peak-picking single trials compares with an averaged voltage within a
time window.
The ST approach used in this study to identify the principal peaks in a
single trial is based on the common method of peak picking to identify
component peaks and latencies in the averaged ERP of each
individual participant in a study. When developing the STP approach,
peak picking at trial level was viewed as a logical extension of
peak picking at the averaged ERP level. That is, any given peak
identified in the averaged ERP is a product of the peaks in the single
trials occurring on or near the time point (latency) of the averaged ERP
component of interest. Furthermore, when the general morphology of the
averaged ERP is captured as successive measures of the major components
(e.g., the N1, P2, N2, P3, N3 peaks),
the time window for finding the
ST peaks of a component of interest is defined by the time points of the
two surrounding components in the averaged ERP (i.e., the latencies of
the preceding component and following component). For example, to obtain
a measurement of the P2 for a single trial one can determine the maximum
voltage between the latencies of the N1 and the N2 components of the
averaged ERP. The latency for this maximum voltage in the single trial
will most likely be close to the latency of the P2 of the averaged ERP
but will vary from trial-to-trial. The use of the prior and post
latencies for determining a peak at the ST level is logical, as every
data point in the single trial contributes to the general morphology of
the data points in the averaged ERP.