Power spectral analysis of heart rate variability(HRV)have been traditionaly employed to determine the variations of the autonomic nervous system (ANS) activity. In practice, short time fourier transform (STFT) and wavelet transform are used as time and frequency localization technique to estimate changes in the power spectra of HRV. The problem, however, is that these methods are limited by their fixed basis function. In this report, we propose an analysis to determine whether or not HRV basis functions trained by independent component analysis (ICA) are useful as alternative method to STFT and wavelet transform to evaluate the ANS variations. Herein we applied learned basis functions to analyze HRV while watching a roller coaster video.Our analysis suggest that basis function learned by ICA is an effective way to explore changes in the low frequency component of HRV power spectra when compared to STFT basis function.