Innovation is the direct product of individual cognitive processes， and we should use the social network to implement innovative behaviors. It has concluded that limited cognitive abilities of individuals would inevitable result in their self-cognition bias in decision-making process， which further lead to the psychological preferences and evaluation deviation of the social network. Social networks not only reflect the interaction of action mode and emotion in the real world， but also reflect the individual cognitive restructuring of social relations. Therefore， although individual overconfidence was once considered to be conducive to innovation activities， but the mechanism of team members’ overconfidence and enhancement of innovation performance could not transplant simply， we need to consider the special role of cognitive social structures in innovation processes． This paper focuses on the impact of overconfidence and cognitive social structures on individual innovation performance through situational experiment， and introduces attributive character of accuracy of cognitive social structures to re-interpret and summarize the analysis logic of cognitive social networks at the psychology cognitive level. Some substantial conclusions could be drawn based on the results. First of all， although the overconfidence in individual decision is considered to be the internal motivation of risk reduction and innovation， but it is likely to be the bottleneck to enhance individual innovation performance in team innovation activities. Team innovation needs knowledge transfer and interaction between individuals frequently， while overconfidence leads to overestimation of the individuals’ ability in terms of knowledge absorption and creation. Even if the overconfidence can promote individual innovation behavior， it is difficult for it to exploit the innovation resources fully， and thus it hinders the refining of creative ideas. Secondly， overconfidence makes individuals overestimate the structural features within a social network， which brings systematic bias to the traditional methods of social network analysis. The higher the degree of individual overconfidence is， the higher the degree of self-perception for individual social network centrality is. Reducing the accuracy of cognitive social structures makes it difficult to achieve higher individual innovation performance simultaneously. Thirdly， self-cognition of social network centrality can partly mediate the effect of overconfidence on innovation performance. Nonetheless， improving the accuracy of cognitive social structures may not be able to promote innovation and enhance individual performance necessarily. Different stages of innovation and varying degrees of innovation will produce different demands on innovative potential and innovative motivation. In some special occasions， optimistic evaluation of embedded structure characteristics may be helpful for individuals to make innovative decisions as quickly as possible. The results of this study can provide an empirical support for project managers with regard to the judgment of team members’ cognitive social structures in the personalized cognitive monitoring， which aims to help employee building objective self-cognitive of social network structure and promote innovation performance．