Does religious symbolism correspond to a particular ‘mode of thought’? In the previous contributions, most authors tried to show, against some classical models, that in terms of cognitive processes it is just not possible to maintain this hypothesis. Whatever the models adopted, the authors tried to show that the formal aspects of religious representations can be found in many other types of contexts: there is no ‘domain-specificity’ of religious symbolism. However, in many cultural contexts religious action and statements are, implicitly or explicitly, ‘marked off from the everyday. Statements are made in a special language, actions are clearly seen to be ritualised, etc. If such special contexts do not correspond to genuine differences in ‘modes of thought’ or cognitive processes, how are they represented? What features or criteria are used to separate contexts? Both contributions in this section aim to examine this question, from the particular viewpoint of acquisition and belief fixation, in the passage from everyday to ritualised situation. Non-ritual contexts are prior, both logically (since rituals are considered ‘special’ against a residual background of unmarked contexts) and in terms of cognitive development (as children only gradually acquire the capacity to tell one type of contexts from the other). Toren focuses on the latter issue, and challenges the common assumption that ‘symbolism’ is entirely separate from, and therefore unconstrained by commonsense knowledge of everyday contexts. This is just not plausible from a developmental viewpoint. Toren takes as an example the way people sit ‘below’ or ‘above’ in Fijian households, depending on gender and other aspects of status.