A Bayesian Time-Course Model for Functional Magnetic Resonance Imaging Data: Comment