It is the task of policy science, as of political theory generally, to recommend a scheme of reason that will presumably result in better public decisions. This may seem profoundly undemocratic, for in a democracy, should not the people be free to think precisely as they please? Yet all political theory intimates an idea of political reason and unreason. The problem then is which ideal of reason to teach in the name of a policy science of democracy. Herein I compare three common schemes of reason: utilitarian calculation, liberal rationalism, and practical reason. My conclusion is that there is no reason to suppose that the capacity for the more public-spirited forms of deliberation is less widely distributed in the population than calculated self-interest, and that either liberal rationalism or practical reason, or some combination of the two, is more compatible with democracy than utilitarian calculation.