Cognitive radio technology increases spectrum utilization by enabling secondary users to opportunistically use the spectrum when primary users are inactive. Secondary users use spectrum sensing to detect the presence of primary users in order to avoid causing harmful interference. To the best of our knowledge, all existing spectrum sensing methods are essentially physical spectrum sensing, in the sense that nodes physically tune their radio to each frequency band to sense the spectrum. In this paper, we propose a complementary approach, virtual spectrum sensing, which achieves the same goal but only senses a very small portion of the spectrum. This approach enables a Distributed Information SHaring (DISH) mechanism, where neighboring users cooperatively share spectrum usage information (obtained from virtual spectrum sensing) with users who need it in decision making. This paper presents an application of DISH to cognitive radio networks. We provide a Cognitive DISH framework which describes guidelines for cognitive radio protocol design based on virtual spectrum sensing and DISH. Under this framework, we design a protocol, VISH-I, and evaluate its performance via simulations. As the number of secondary users increases, the interference caused to primary users results in only 5% performance degradation, but the overall channel utilization is increased by 87-203%. In addition, to demonstrate that virtual sensing is complementary to physical sensing, we design a hybrid spectrum sensing protocol, VISH-II, which improves performance by 7-50% over VISH-I.