This study presents the outline of a model for collective phenomena. A symmetry-breaking model combines a number of well-established social psychology hypotheses with recent concepts of statistical physics. Specifically we start out from the regularities obtained in studies on the polarization of attitudes and decisions. From a strictly logical point of view, it is immediately clear that aggregation effects must be analysed separately from group effects as such. The conceptual analysis of the assumed mechanisms reveals that when we deal with phenomena that have until now been designated as polarization phenomena, we are faced not with a single phenomenon, as was believed hitherto, but with a whole class of phenomena. For this reason it would be appropriate to deal with them differentially both from an empirical and from a theoretical point of view. It is possible to show, moreover, that in principle polarization is a direct function of interaction and, beyond a critical threshold an inverse function of the differentiation between group members. A certain number of verifiable conjectures are presented on the basis of physio-mathematical-psychological considerations. It is to be hoped that these theoretical outlines will make it possible to give a new lease on life to a field of research that has established solid facts, but that became trapped in a dead-end road, for lack of a sufficiently broad analysis.