Abstract: Wrist-worn wearable devices provide rich sets of pulsatile physiological data under various modalities and circumstances. An unexploited capability is that the pulsatile physiological time series collected by wrist-worn wearable devices can be used for recovering internal brain dynamics. We present two design classes of closed-loop wearable brain-machine interface architectures related to cognitive stress for tracking arousal and fatigue states. The methods are validated by analyzing experimental electrodermal activity and cortisol data as well as simulation studies in the context of cognitive-stress-related arousal and fatigue. Results demonstrate a promising approach for tracking and regulating neurocognitive stress through wearable devices. Since wearable devices can be used conveniently in one's daily life, wearable brain-machine interface architectures have a great potential to monitor and regulate one's neurocognitive stress seamlessly in real-world situations.