Certain discursive organizations of objects, concepts, and enunciative modalities give rise to 'themes' or 'theories' (the former denoting less 'coherence, rigour, and stability' than the latter). Foucault calls these themes and theories 'strategies.' He cannot say as much about the analysis of the formation of strategies in history as he was able to about the analysis of the first three formations, because his attention in his three previous books was directed primarily toward these first three. He will, however, indicate some directions for research.
First, we must determine the 'points of diffraction' of a discourse. These points occur when two incompatible objects, concepts, or types of enunciation have the same conditions of emergence and try to occupy the same discursive space (an 'either or' situation). Often, whole discursive 'sub-groups' are attached to these incompatible elements. Second, the factors that determine which these conflicting possibilities actually becomes part of the discourse often exist outside of the discourse in question. One must therefore look for these factors in the functioning of the 'economy of the discursive constellation' to which the discourse belongs, examining modifications in principles of exclusion and choice. Finally, the determination of these theoretical choices is also made possible by a 'field of non-discursive practices,' a field of authority governed in part by restrictions on who can say what where, who can spend what where, etc. This field of authority is also determined partly by the place of desire in relation to the discourse, which may serve as a location for fantasy, the forbidden, satisfaction, etc. These are not factors hidden within the discourse or external to it; they serve as some of its 'formative elements.'
Just as objects, enunciated statements, and concepts each had a realm to which they must not be related (things and words, knowledge and the psychological subject, and pure ideas, respectively), so theoretical choices must not be related to the development of a fundamental 'project' of the discourse.
Chapter 7: Remarks and Consequences
Foucault now pauses to consider what he's done in the foregoing chapters. The most urgent issue concerns the replacement of the traditional, obvious unities of discourse with these much more complex, even 'dubious' regroupings he has called 'formations.' Where does this get us? Do formations really allow us to perceive any kind of individualized wholes of discourse? Or are we always left with what we started with, a dispersion of elements, an 'irreducible multiplicity'?
First of all, we should consider that the four levels of formation described here are not independent; each limits the possible relations of the others. Thus, we are already dealing with a restricted discursive field, where not just anything is possible in each level. Second, the discursive formation exists nowhere except in the discourse itself. Therefore, it is a formation bound up in time, a 'schema of correspondence between several temporal series.' The rules that individualize a given formation (like psychiatry) can remain constant even as new objects, enunicative modes, concepts, and strategies appear according to that rule. On the other hand, the whole formation is slowly being affected by its relations with other discursive formations.