Abstract
This article first tackles the issue of defining what counts as a religious belief, and shows why obtaining such a definition opens the way to discovering a deeper level of interaction between divinity beliefs and the scientific enterprise than the prevailing views of the science/religion relation allow for. This deeper level of interaction is illustrated by applying it to twentiethcentury atomic physics. It is then shown why this level of interaction implies a distinctive anti-reductionist perspective from which theists should do science, a perspective in which belief in God acts as a regulative presupposition. Finally, reduction as a strategy for explanation is critiqued and found bankrupt.