The starting point of the discussion is provided by Luhmann's concept of society as a social system as well as his concept of systems media. It puts forth the argument that a theory of society cannot be reduced to an analysis of society as a social system in which communications are considered to be the only legitimate unit of sociological analysis. A fully–fledged theory of society should be able to explain society as both interaction and communication. Moreover, the creation of such a theory will involve the observation of society from both sides – from the inside (where Luhmann conducted his observations) as well as the outside (where interaction, not action, is the basic unit of analysis).
The author argues that Luhmann was fully aware of the one–sided character of his theory of society. This is evident from his Observations on Modernity, where he concluded that an objective observation of society from both inside and outside is theoretically inconceivable. His attempt to overcome the problem with the help of Spencer Brown's Laws of Form was also unsuccessful. Because crossing and re–crossing the systems border takes time, it is possible only to observe the inside of a system in T1 and its outside in T2. Time thus emerged as the greatest obstacle to a simultaneous observation of modern society as a whole.
The present discussion shifts the focus away from the Habermas–Luhmann controversy, which drags sociology into a battle with social philosophy that cannot be won, to the internal boundary that divides macro– from micro–sociology. This boundary is identified as the boundary that divides Luhmann's systems theory from George Herbert Mead's symbolic interactionist theory. The author's conviction is that previous difficulties with viewing Mead's theory as the "other" of Luhmann's have very much been the product of historical circumstances. Mead's work was labelled "social psychology" after his death and buried under a distinct approach that Blumer termed "Symbolic Interactionism." Only recently have Meadean scholars begun to see the "irony" in this historical development (Lonnie Athens). The essence of this irony is that while Mead's work apparently suggested a particular conception of society, whereby it was viewed as consisting of symbolic interaction between role players who take each others' roles, it came to be known and appreciated in sociology solely for its conception of self. It thus remained for an entire century under the aegis of sociological social psychology, which took little interest in what society really was.
The other important concept in Luhmann's theory, the "medium", seems to be even more problematic. At the very beginning of his theoretical undertaking Luhmann encountered an insoluble theoretical problem, namely, that sociology, as he could tell, had no concept of medium. Although he found motivation for his explorations in Parsons' analysis of social media (such as language, money, and even power), he found there no sociological theory of medium appropriate for macro–sociological use. The paper suggests that a theory of the medium can in principle be neither discerned, nor fashioned from a purely sociological perspective. The claim is that the theory of the medium is a socio-psychological issue. The present discussion proposes that Luhmann's theoretical bias or "blind spot" as a macro–sociologist was in fact the micro–sociological theoretical world.
Luhmann apparently shared the bias against micro-sociology of his teacher Talcott Parsons. His comments on Symbolic Interactionism are superficial, ambiguous, and infrequent. Luhmann followed the Parsonian strategy of borrowing from other disciplines what sociology appeared to lack and turned to the rival tradition of psychological social psychology and to the work of Fritz Heider, who worked in the gestalt tradition. Nevertheless, the claim is that Luhmann's approach is not quite Parsonian. He does not simply borrow ideas from other disciplines. He rather borrows theoretical forms that are then "customized" in light of sociological needs and imposed upon sociological matter.
Finally, a more thorough examination of Symbolic Interactionism reveals that it has never been a unitary approach. Blumer, who coined the term and subsumed under it the views of a great variety of theoretical perspectives because of their "general similarity," appointed Mead's social–psychological views as representative of the approach as a whole. The present paper argues, in contrast, that two different versions of social psychology exist within Symbolic Interactionism, and that they are bound to two different versions of society and to two different versions of social selves. These are the versions that were elaborated in the works of Cooley and Mead, and the distinction between them is elaborated in terms of two different understandings of what symbolic interaction is. It is argued that Cooley's theory, which itself is based upon the communication theory of Gabriel Tardé, is a good starting point for a theory of social medium, something that could not be found by Luhmann. A radically different description of medium can be inferred from Cooley's perspective. For example, following Cooley, one may argue that the medium has an agency, and that it is the medium which, through interaction, constructs its own form, not just the loosely connected elements upon which the system imposes its own form, as Luhmann suggests.