In this short editorial, we outline the nature of belief as it is understood in cognitive science (in particular, in computational neuroscience), wherein beliefs are understood as representations of the states of the world. We then introduce active inference: a process theory that formalises how organisms update their beliefs about the sensed world, and how they operate in the world to minimise the uncertainty (or entropy) of beliefs about the causes of their observations. Finally, we examine how active inference manifests across scales, and the different ways in which belief dynamics emerge over scales, providing an explanation for the wide variety of socio-political phenomena and psychopathology.