Orndorff and his colleagues [1] suggested that if a neural activity is considered a treatment variable instead of outcome, it widens the scope of research and has a specific implication for social neuroscience. Given this, the empirical evidence is collected and analyzed where neural activity as self-manipulation design through neurofeedback training specifically for social cognition deficit is done. The objective of the present article is to provide a systematic review of 1) how NFT is utilized to treat social cognitive deficits, 2) how NFT is utilized to target Social Cognition Deficit in ASD, 3) examining the directions, strengths, and quality of evidence to support the use of NFT for ASD. The databases for studies were searched in PubMed, MEDLINE, EMBASE, Springer, Science Direct, Psychinfo, and Google Scholar, using combinations of the following keywords: ‘Neurofeedback,’ ‘Autism Spectrum Disorder,’ ‘Mu Rhythm’ and ‘Social Cognition.’ Studies were eligible for inclusion if they were specific to 1) autistic and typically developed population, 2) intervention study, 3) Delivered by NFT, 4) participants showed social cognitive deficit and/or improvement. Total one eighty-seven studies were found of key interest; out of which 17 studies were eligible for inclusion in this review. All studies reported the improvement in different domains of social cognition and were moderately methodologically sound. Eleven out of seventeen studies satisfied the trainability and interpretability criteria suggested by Zoefel and his colleagues [2]. The conclusion from the present review is in line with comments of Marzbani and colleagues [3] that, ‘current research does not provide sufficient conclusive results about its efficacy.’ The patterns and directions concluded from studies related to protocol, methodology and results are discussed in detail in the present review. © 2020