Functional near-infrared spectroscopy, a neuroimaging tool used to measure brain activity, is associated with two different types of artifacts: (a) task-induced superficial-tissue hemodynamic artifacts derived from the scalp; and (b) motion artifacts caused by head motions. Recently, a simple and easy-to-use method, based on a general linear model incorporating superficial hemodynamics model estimated from short-probe distance channels using principal component analysis, was proposed to remove task-induced superficial-tissue artifacts. In the present study, we examined the effectiveness of this method in removing task-induced and head motion-induced superficial-tissue hemodynamics. Thus, we conducted a motor experiment where participants were asked to perform grasping movements. During some sessions, head motions were generated in order to introduce motion artifacts. Although the removal of motion artifacts was not perfect, we found that analyses including the first and second principal component (estimated from short-channels) showed a tendency to provide accurate detection of brain activity. This finding demonstrated the possibility of conducting effective analysis of functional near-infrared spectroscopy using general linear model and short-channels.