Componential methods such as PCA and ICA estimate the individual spatial and temporal components for a given subject and a given condition separately and do not allow the simultaneous comparison of ERP components across subjects and conditions (Wang et al. 2000). Researchers therefore developed trilinear modeling, a novel method for estimating a set of spatial components (brain maps) and temporal components (waveforms) of time-locked brain potentials across subject groups and task conditions (Wang et al. 2000). Trilinear modeling is one member of a family of modeling techniques that extends two-dimensional linear modeling to multidimensional modeling, in general known as N-way modeling. Trilinear modeling is based on the topographic component model (TCM) (Mocks 1988), which models brain potentials in a trilinear form. The trilinear approach builds on singular value decomposition (SVD) and extends the TCM mainly by replacing the diagonal amplitude matrix by a general loading matrix and by allowing the number of spatial and temporal components to be different (Wang et al. 2000). Thus, the trilinear model has the advantages of both SVD and TCM methods. The trilinear components are
