In the present age, the progress of our knowledge concerning the forms of religious membership depend less on more precise instruments for measuring behaviour and explicit beliefs than on hypotheses derived from sociology of knowledge and of culture. This implies in its turn the in vention of new techniques, where the possibilities of interviewing are criticized. We were faced with the question in our work with F. Boulard: comparing the results of the census of Sunday Mass attendance in the majority of French towns. This change in scale in relation to monographic analyses has brought to light a link between religious observance and the cultural structure of the group, that is to say collective cultural patterns pro gressively formed in the course of history and transmitted in a quasi- hereditary fashion. This leads us to propose the following hypotheses: religious behaviour is an expression in daily life, of implicit cultural patterns, which direct the believer to diverse behaviour and beliefs. These cultural patterns operate even if they are unknown, They are transmitted independently. Thus the social efficacy of religion would be due to its insertion in latent cultural structures and to unconscious transmission. A religious crisis would be more serious if it results from an evolution of these latent cul tural patterns. This affects the conception of the person who through an original work, tries to express, and even to invent in an unconscious way, these collective patterns. In order to verify these hypotheses, it seems important to detect several levels of interpersonal communication and examine the extent to which they are in continuity or not. Such was our aim in two research studies. One, entitled 'Analysis of the ideology of students at the University of Louvain', attemps to compare religious membership arising from criteria defined by authority, the church as a reference group, religious and ideological evaluation of oneself and one's environment, various stereo types of social life, as also some future projects dealing with criteria around which worlds of plausibility are organized. The other study at tempts to compare three types of languages: the results of a classical type of interview; a projective test (the construction of an ideal town) analysed by a structural method; the commentary made by the subject on his projection.