A tutorial on the use of temporal principal component analysis in developmental ERP research – Opportunities and challenges

Abstract

Developmental researchers are often interested in event-related potentials (ERPs). Data-analytic approaches based on the observed ERP suffer from major problems such as arbitrary definition of analysis time windows and regions of interest and the observed ERP being a mixture of latent underlying components. Temporal principal component analysis (PCA) can reduce these problems. However, its application in developmental research comes with the unique challenge that the component structure differs between age groups (so-called measurement non-invariance). Separate PCAs for the groups can cope with this challenge. We demonstrate how to make results from separate PCAs accessible for inferential statistics by re-scaling to original units. This tutorial enables readers with a focus on developmental research to conduct a PCA-based ERP analysis of amplitude differences. We explain the benefits of a PCA-based approach, introduce the PCA model and demonstrate its application to a developmental research question using real-data from a child and an adult group (code and data openly available). Finally, we discuss how to cope with typical challenges during the analysis and name potential limitations such as suboptimal decomposition results, data-driven analysis decisions and latency shifts.

Identifying Reliable Neural Components and P300 Latency using Correlated Component Analysis

“…Recorded EEG channels are usually noisy. Techniques such as principal component analysis (PCA) [34], [35] and independent component analysis (ICA) [36], [37] are often used to remove noise and extract the ERP signals. In this work, we used CorrCA to nd the components having maximal correlation across the target trials cohort.…”

Intention-based predictive information modulates auditory deviance processing

“…In other words, the identification of ERP components in the measured ERPs is obscured because the measured ERPs are a mixture of latent underlying (sub-) components. Spatial and temporal overlap considerably biases the observed component peaks typically used to identify and label components ( Scharf et al, 2022 ). Moreover, the practice of determining time windows for the respective components based on (peaks in) the observed ERP frequently suffers from the relatively arbitrary definition of time windows and double dipping ( Kriegeskorte et al, 2009 ).…”

“…Moreover, the practice of determining time windows for the respective components based on (peaks in) the observed ERP frequently suffers from the relatively arbitrary definition of time windows and double dipping ( Kriegeskorte et al, 2009 ). Temporal PCA largely reduces these problems (e.g., Dien, 2012 ; Scharf et al, 2022 ). For that reason, we used temporal PCA to delineate the components in a straight-forward, data driven approach.…”

“…We conducted temporal principal component analysis (PCA) on the individual average ERP data of all channels and stimulus types using the tutorial code provided by Scharf et al (2022) . PCA was computed using Promax rotation (κ = 3) with a covariance relationship matrix (preferable over correlation matrix for ERP analyses as all sampling points are measured on the same scale; for discussion see, Dien et al, 2005 ; Scharf et al, 2022 ) and Kaiser weighting (to ensure that each variable has equal influence on the rotation process and therefore prevent that large factors dominate the results of the rotation step; for discussion see, Dien et al, 2005 ; Scharf et al, 2022 ). The number of components to be retained was determined using Horn’s parallel test.…”

Single-trial-based temporal principal component analysis on extracting event-related potentials of interest for an individual subject

“…Due to the volume conduction, ERP components overlap in temporal and spatial domain to some extent. Thus, the factors are required to be correlated substantially (Dien, 1998;Scharf et al, 2022). Besides, the results of actual and stimulated ERP datasets indicated that Promax rotation showed more accurate results than Varimax rotation (Dien et al, 2005;Dien, 1998;Dien et al, 2003).…”