Foucault now attempts to give some 'real content' to the notion of 'rules of formation.' He will begin by looking at the formation of objects, using the example of psychopathology from the late nineteenth century onwards (whose objects seem range from major ones like 'madness' to minor, more specific ones like 'sexual aberrations and disturbances' or 'criminality'). The objects of psychopathology arise in different ways, and some get completely erased; all are transformed over time. What rules govern these processes?
First, we can consider 'surfaces of emergence,' the fields in which an object first arises. These can be pre- existing fields like family, social group, work situation, etc., each of which is normative to some degree, each of which has developed a 'margin of tolerance' that roughly defines the field by what it rejects (in the present example, by what it deems 'mad'). There are also markedly new surfaces that begin to emerge in the late nineteenth century, such as art (with a new, strict normativity), sexuality (as an observable field of possible deviations), and penality (in which madness and crime become linked for the first time). All of these serve as surfaces of emergence for the objects of psychopathology; they are fields of 'initial differentiation,' whose 'distances, discontinuities, and thresholds' allow psychiatric discourse to define what it is talking about (thereby creating apparently definite objects of discourse).
Second, the 'authorities of delimitation' must be considered. Who had the authority to 'delimit, designate, name, and define' objects like madness? What was the structure of their power (both in its organization and in how it was publicly perceived), and what were the processes by which they adjudicated the limits of a given object? Finally, we must analyze 'grids of specification,' the systems by which madnesses were described, separated, and classified (for the nineteenth century, Foucault lists 'the soul the body the life and history of individuals [and] the interplays of neuropsychological correlations,' each system with its own organizing principles).
But this picture of the emergence of discursive objects is still only partial, because it fails to address the complexity of the relations between the object and its plane(s) of emergence on the one hand, and between the various planes on the other; none of these are truly separate from each other. To really examine the emergence of discursive objects, our focal point must be not the individual planes of emergence, but their overlaps, tensions, and relations as they give rise to discursive objects. Similarly, we will not find certain 'privileged objects' that define a discourse, but rather a dispersal of many objects. It is thus the complex formation of this dispersed field of objects that characterizes a given discourse (here, psychopathology).
This analysis generates four 'remarks and consequences:' 1. Objects do not pre- exist their emergence under certain complex, relational, discursive conditions. 2. The object is therefore not defined by its internal, conceptual nature, but by its exterior relations, its triangulation or juxtaposition with other objects in a 'field of exteriority.' 3. It becomes important to separate properly 'discursive relations' from 'primary' and 'secondary' relations within a discourse. An example of a primary relation might be the relation between the nineteenth-century bourgeoisie and the judicial system; a secondary one would be a nineteenth-century psychiatrist writing about the relationship between criminality and psychology. Neither level will consistently show us how a discursive object is formed; such levels cannot always 'superposed' on the relations responsible for the formation of objects. These latter relations are 'properly discursive' ones, and they interlock with the other levels in complex ways. 4. Discursive relations are thus not 'within' discourse itself, but neither are they 'exterior' to it; they take place 'at the limit of discourse.'
In this analysis, then, psychopathology as a provisional, reflexive category has given way to a dispersal of objects and discursive relations that cover more ground, and in greater complexity. We have been able to see psychopathology 'as a discipline in a constant state of renewal.' That discipline's constancy lies neither in its objects nor the ways in which those objects emerged or were characterized, but rather 'the relation between the surfaces on which they appear.'
In closing, Foucault emphasizes again that this method is not linguistic, not an attempt to understand the 'meaning' of terms like 'madness.' Discourse is not a matter of words, and neither does it lie in the meeting of words and things, lexicon and experience. Discourses are not groups of signs that refer to contents or to 'dumb reality,' but are rather 'practices that systematically form the objects of which they speak.' They are also 'more' than this; this 'more' will be addressed further on.
After rejecting 'hypotheses' in the chapter on 'The Unities of Discourse,' Foucault begins to build his own theory. The first hypothesis in the chapter 'The Unities of Discourse' is that a given discourse can be individualized by virtue of the 'object' of knowledge that it uniformly addresses (the pervasive example here is 'madness,' the supposedly constant object of psychopathology). The problem with that hypothesis was that the 'object' refused to remain a constant; not only did it split into multiple, shifting objects, but its very existence was found to be less a matter of the object itself than of its exterior relations. Now Foucault asks what the object of a discourse looks like in his method, reconstituted from the ground up. The provisional answers offered here will also tell us something about the nature of the discourse whose object of knowledge is in question; in fact, the two are inseparable.
The three initial answers Foucault gives to the question of what 'rules' the existence of objects of discourse are pleasingly practical after the speculative deconstructions of the previous chapters. We must examine the contexts in which the object is initially situated (the 'surfaces of emergence'), the authorities by which it is delimited, and the systems by which it is structured and differentiated from other objects ('grids of specification'). (Foucault's examples in this breakdown are worth reading; they give a sense of how one would actually go about investigating these three factors.) Despite the practicality of these levels of analysis, Foucault is as rigorous as ever in never allowing us to rest on familiar categories. The surfaces of emergence for madness alone range far from what we know as psychopathology: family, religion, sexuality, labor, and penal policy are all central, for starters. Furthermore, 'madness' emerges from these fields not simply as an invention or a hypothesis, but in complex processes of exclusion. These processes have everything to do with power relations (a subject to which Foucault will devote a great deal of attention in other works), and in turn with authority, the second of the points of approach listed here.
Foucault uses the example that religion was 'probably' a surface for the emergence of madness as an object of knowledge in the nineteenth century because it is a normative community that depends in part on the articulation of what is abnormal (here, madness). The 'authoritative' function comes in here, with church authorities defining, say, what is a religious vision and what is a hallucination. In so doing, they transfer instances of madness to the field of medicine for explanation and treatment. Medicine then further classifies and defines madness by virtue of interlocking systems of classification. Throughout all this, there is no constant called 'madness' that gets authoritatively defined, thrown out, diagnosed, and treated. Each of these processes plays a role in triangulating the thing called madness, and madness arises as an object only in the complex relation between the steps.
But this is only an introduction to the complexities with which Foucault freights the concept of the discursive object. He also notes explicitly that relationships like those described just now are not in and of themselves the relations that fully account for discursive objects. Our example includes only 'primary relations.' 'Secondary relations' would include reflections, at the time, on the relationship between the family and the church in the identification of madness. Even here, however, we have not reached the level of relations in which objects fully emerge. That level is the level of the 'discursive relation,' which depends on primary and secondary relations but reveals interactions implicit in neither of them.
In this tri-level distinction, Foucault is pointing out that the conditions of the emergence of an object are usually not apparent at the time; these conditions emerge from a bigger picture. But there is a more significant point in the background, here, namely that discursive relations establish the conditions of possibility for statements about a given object of knowledge. Statements made at the level of primary or secondary relations speak according to these conditions of possibility, but are generally not aware of them. The object of discourse is what is 'given to the speaking subject'; the network of relations in which that object appears make it possible for the subject to say just what they say (rather than anything at all). In this sense, the Foucauldian historian has a particular kind of subjective privilege with regard to the past: only he or she can trace relations at the discursive level.
Foucault rejects a linguistic approach toward the end of this chapter. He rejects the idea that his method seeks the 'meaning' of terms like madness. This argument is largely explained by Foucault's constant insistence on the relational nature of discourse in its relevance for the historical project. To consider madness as a word (a sign) with a particular meaning, a particular concept, would be to miss the work that Foucault is trying to do in showing how madness appears and operates historically precisely as an effect of relations between groups of statements. Foucault will consider this distinction in further depth. For now, he says only: 'Of course, discourses are composed of signs; but what they do is more than use these signs to designate things. It is this more that renders them irreducible to the language and to speech. It is this 'more' that we must reveal and describe.'