Cortical PCA: Exploring Potential Tempo and Transient Locking to Musical Stimuli

Abstract

We explore whether dynamic principal component analysis (PCA) of cortical EEG amplitude or power envelopes reveals tempo and/or transient locking to musical stimuli. Using log-transformed, z-scored amplitude or power envelopes across a range of frequency bands, we project whole-cortex EEG dynamics into a low-dimensional PCA space and track trajectories over time. These trajectories exhibit stable rotations, song-specific attractors (with occasional overlap between songs), and orbit structures that vary systematically with PCA parameters (e.g., window size, Gaussian taper shape) and frequency band. We compared the tempo of several songs with the natural oscillatory tempo of PCA ring trajectories in the first three dimensions of PCA space, finding that PCA ring rate held steady around 5.14 Hz invariant to song tempo. We also explored whether transients align with sharp directional changes in PCA space, considering a wide range of spectral bands.