EEG neurofeedback can help individuals learn to self-regulate their brain activities and has been shown benefits in health and performance enhancement. This thesis aims to first utilize neurofeedback training for improving short term memory and peripheral vision, which are indispensable brain functions but find no successful improvement records in the literature. Meanwhile, since EEG learning (i.e. selfregulation of EEG during neurofeedback) that has crucial mediation link with neurofeedback efficacy has been reported large individual difference and even nonsuccessful cases, the second objective of this thesis is to find out predictors for EEG learning. Firstly, individual alpha neurofeedback protocol was proposed to improve short term memory. Sixteen subjects received 20 sessions of neurofeedback aiming to upregulate the individual alpha amplitude. Compared to the non-neurofeedback control group, the neurofeedback group showed significant larger enhancement in short term memory. Importantly, the improvement of short term memory was positively correlated with the increase of the individual upper alpha during training, confirming the effectiveness of individual alpha neurofeedback for short term memory. Secondly, peripheral visual performance was found positively related to the alpha amplitude during visual task. Based on this relation, individual alpha neurofeedback was proposed for improving peripheral visual performance. Thirteen subjects performed 20 sessions of neurofeedback, leading to successful enhancement of the alpha amplitude during the visual task and also the peripheral visual performance compared to the non-neurofeedback control group. These findings contribute original iv evidence of improving peripheral visual performance using individual alpha neurofeedback. Finally, the predictors of EEG learning in both alpha and beta/theta ratio neurofeedback were investigated. It was found that the resting alpha amplitude was the predictor of EEG learning in the alpha neurofeedback while the beta amplitudes in resting and initial training were the predictors of EEG learning in the beta/theta ratio neurofeedback. Importantly, higher predictor value was associated with better EEG learning in both protocols. These findings imply the significance of resting EEG features in self-regulation of EEG by neurofeedback, prevent potential frustration and time consuming training sessions for the individuals with low EEG learning, help modify the training protocol accordingly and understand the mechanisms of neurofeedback.