A young Hui person says, “I’m not really religious,” then refuses pork because “I’m Hui.” This essay takes that ordinary moment as a philosophical problem. I argue that it cannot be classified by any definition that makes religious practice depend on a single essential feature, especially explicit belief. A polythetic account of religion handles it better, but, as I show at the end of this part, even that account requires revision.
4.1 The problem with essentialist definitions
The most intuitive definition of a religious practice is belief-centred: a practice is religious if its practitioner performs it because of belief in God or the sacred. This works for clear cases, such as a Muslim performing daily prayer as an obligation to Allah. It fails, however, for the large class of practices that persist after explicit belief weakens.
Alternative single-feature definitions fare no better. A definition based on belief in a personal God excludes some Buddhist traditions. One based on supernatural content cannot explain secular commitments that organize a person’s moral life in religious ways. One based on intensity includes too much, since people are intensely committed to many things we do not call religious. Each definition generates counterexamples, which suggests the problem lies with the search for a single essence itself.
4.2 A polythetic alternative
In earlier work, I drew on Wittgenstein’s notion of family resemblance (Philosophical Investigations, §66–67) to argue that “religion,” like “game,” names a category whose members share overlapping features rather than one common property. On this polythetic view, a belief is religious to the extent that it exhibits some combination of three dimensions:
Structural: it organizes the believer’s other commitments.
Relational: it is held within a community that treats it as authoritative.
Classificatory: it is recognized by a religious tradition or makes a characteristically religious claim.
No single dimension is necessary. A belief can be religious by satisfying several dimensions without satisfying all of them.
4.3 Test case: four refusals
Hui dietary practice provides a useful test of this model. Research on qingzhen among Chinese Muslims shows that food practice expresses ethnicity, community membership, trust and social boundaries alongside religious observance. The same outward act can therefore rest on very different grounds. Consider four Hui individuals who each refuse pork:
A refuses because the Qur’an prohibits it.
B refuses because eating it would betray what her parents passed down.
C refuses because “we have never eaten it in my family.”
D does not experience refusal as a choice; pork simply does not register as food.
The three-dimensional model classifies the first three cases well. A’s refusal is structural, since explicit belief governs the act. B’s is relational, since its authority derives from family and community. C’s is classificatory, since the act is followed as part of a recognized tradition.
D cannot be classified. D holds no belief about pork, invokes no authority, and appeals to no tradition. Yet D’s refusal is plainly continuous with A’s: it has the same origin, the same content and the same social meaning. A belief-centred definition would declare D secular. The polythetic model, as formulated, has no category for D at all.
4.4 The inverse case
My fieldwork in Ningxia produced a case that runs in the opposite direction. Hui families I interviewed in Yinchuan and Tongxin told me that some non-Muslim neighbors had come to prefer halal food, not for religious reasons but because they considered it cleaner and healthier.
This case satisfies none of the three dimensions from the consumer’s standpoint: there is no belief, no religious community and no sense of following a tradition. It is, correctly, not a religious act. Yet its appeal derives from qualities, cleanliness, care and trustworthiness, that were shaped by Islamic purity law. This shows that a practice’s religious origin is not sufficient to make an act religious. That is an important limit, and I return to it in Part 6.
4.5 Conclusion
Taken together, the two cases mark the boundaries of the problem. The non-Muslim consumer shows that religious origin alone does not make a practice religious. D shows that the absence of belief, authority and tradition does not make a practice secular either. What D retains is a boundary held in the body. Part 5 argues that this embodied dimension is real, distinct and irreducible to the other three.
