Uncertainty and divergent expectations over global warming make it difficult to achieve a majority coalition supporting carbon taxes. We explore a state-contingent approach based on an updating rule that automatically assimilates new information rather than a pre-specified tax path. Agents form expectations which imply that the tax sequence correlates with their preferred price trajectory. We show that whereas greater variance in beliefs about future global warming undermines support for a static policy, the state-contingent proposal attracts majority support irrespective of the divergence of views, and even has robustness properties to strategic voting by dishonest agents.