In the vast anthropological literature devoted to religious behaviour and discourse, one particular domain is relatively neglected: that of the psychological, more precisely of the cognitive dimensions of religious behaviour. This is particularly surprising now, in view of the enormous development and complexification of ‘cognitive’ science, that is, the heterogeneous cross-breeding of theories and hypotheses put forward in linguistics, psychology, artificial intelligence and philosophy. The aim of this introductory essay is both to account for this state of affairs and to suggest some directions of investigation in the cognitive study of religious symbolism.

The term ‘symbolism’ is particularly difficult to define. As Sperber points out (1975b: 2), the domain is construed either as the mental minus the rational or the semiotic minus language. A theoretical difficulty is thus reified as a real domain of thought and action, which justifies Gellner's wry comment (1987: 163), that ‘in social anthropology … if a native says something sensible it is primitive technology, but if it sounds very odd then it is symbolic’. I will deal presently with the important problems involved in delineating the domain of ‘religious symbolism’, more specifically with the ambiguities of the very notion of‘symbolism’. At this stage, however, the term should be understood, in a fairly straightforward way, to denote the ideas, beliefs, actions and interaction patterns which concern extra-human entities and processes. In this domain of cultural interaction, certain conceptual domains, concerning for example, ancestors, spirits, gods, sacrifice, etc., stand out as especially difficult to describe in a precise and explanatory way.