Religion is one of those social realities that can be characterized in many different ways. Some critics take a realist approach, maintaining that there really is such a thing as religion, and that the notion of a religious entity names a distinct class of phenomena that reflects a set of shared features in social life. Other critics argue that it is a mistake to focus on beliefs, and that the concept of religion should be analyzed in terms of its social practices and institutions, or even more broadly as a set of values.
Many approaches to the study of religion use a combination of these two perspectives. For example, Edward Burnett Tylor argued that to define religion narrowly as belief in spiritual beings would exclude many societies from the category of “religious”. Others have used a more open polythetic approach, recognizing that many properties are common or typical of religions, but that these can also be found in other social phenomena and so do not constitute the essence of religion.
The most recent discussion of this issue focuses on the problem of identifying what makes something a religion, based on an attribution theory approach to social kinds (Smith 2008). Smith suggests that there are secondary features and powers that are important to any religious phenomenon, including identity, community, meaning, expression and experience, and social control, but that these do not make up the essence of religion. Instead, he claims that the essence of religion is self-perpetuating formative causal powers that generate and strengthen the religions they create and sustain.