[Purpose/significance] Professional virtual communities are important platform for sharing and creation of subject knowledge. It is essential to sustainable development for communities to make lurkers change into contributors, improve continuous participation and reduce the loss rate of users. [Method/process] Based on the theory of social learning, this paper discusses the influence of two social learning processes, observation learning and reinforcement learning, on knowledge contribution behavior of lurkers and contributors in initial and continuous participation stages respectively, and forms theoretical model from environment stimulation (observational learning, reinforcement learning), organism cognition

(sel-fefficacy, outcome expectation) to behavior response (initial contribution, continuous con? tribution) . [Result/conclusion] Empirical analysis based on the data collection from bbs.pinggu shows that: in the stage of initial participation, observation learning has an important impact on lurkers’organism cognition, and indirectly affects the initial knowledge contribution behavior through self-efficacy and outcome expectation; in the stage of continuous participation, the effect of observational learning on organism cognition of contributors is significantly reduced, only has indirect impact on contribution behavior through outcome expectation, and it can also partly influence continuous knowledge contribution through the mediating effect of self-efficacy and outcome expectation; compared with out? come expectation, self-efficacy has a stronger influence on continuous knowledge contribution for contributors.