This overly long annotation diverges a bit from the reasoning behind this quote's inclusion here... But I still think what I have to say is relevant to the "explanatory pluralism" and "Thinking across Scales" design logics, and to PECE more generally. And it was a place for me to work out some of my thoughts on this matter... So I went ahead with it.
What I find interesting about this quote by Fischer is that he seems to be tracking the increased granularity of the analysis of the epistemic infrastructure of meaning. By that I mean, he begins with the behaviorists who take words/symbols as "unproblematic tokens." That is, the units of analysis were the words/symbols themselves, without any concern for the meaningful components and relations between these components of which they were composed. Thus, the scale of analysis, and the scale at which cultural patterns could be revealed took place at the level of sets of words, utterances, sentences, etc. And taking words/symbols as one's basic unit of analysis promotes a linear-code type model of culture, the end goal of which was to line up the sequences of behaviors to reveal its patterns.
Symbolic anthropologists, by contrast, took a polysemic approach to the symbol. That is, they saw symbols, not as parts, but as heterogeneous and complex wholes, as being made up of parts themselves. Thus, I would argue, the symbolists didn't actually take "symbols" as their basic unit of analysis, but rather "meanings." And I'll attempt to justify this by citing Fischer's point that the symbolists recognized how any communicative act produces more "meaning" than either the sender or the receiver intends or comprehends. Thus, the granularity here is "turned up a notch," if you'll pardon the phrase, because we've begun to fracture the "word" or "symbol" into something more complex than its definition. But meaning is a rather obtuse and ambiguous unit... Thus, taking meaning as one's unit of analysis does not accord for the same kind of linear-code model of culture as did the behaviorist. It requires a physics, rather than a genetics, something more than a code that can appreciate and distinguish intensities. Thus, as Fischer also notes, the symbolists tended to think in terms of nested centers and peripheries (like the black holes at the center of a galaxy), that scale up all the way from the symbol (as all of the possible meanings of a given symbol pivot around its core meaning or "essence") to clusters of core symbols (as some symbols hold more weight in the culture than others) to the culture itself (which orbits around those clusters of core symbols).
The post/structuralists, then, perform yet another "turn of the screw" by taking, not meaning, but "relations" as their basic unit. This, as Fischer states, effectively "decomposes" the symbol and the symbolic anthropologist's gravity-based model of meaning along with it. Instead, by plumbing the quantum depths of the symbol and finding no essence at the its center, and rather a system of differences, the symbol and its meaning are effectively turned inside-out. "Culture," for the structuralists, is (like Chomsky's generative grammar) an abstract machine - structured by delimited set of fixed relations but, from within those relations, one that is capable of producing near (if not) endless variation. And then, for the post-structuralist, even this (or any other) characterization of the "machine" would be the machine's attempt to characterize itself, such that the machine producing the very distinction between itself and its variations... falls apart. And thus analysis, for post-structuralists, involves a precise characterization (if not production) of the process by which that machine falls apart.
So, if you're wondering why I practically just rewrote exactly what Fischer wrote... it's because I wanted to highlight the point that, within the category of meaning, the units of analysis play a considerable in determining the epistemic relations of the "abstract machines" that produce meaning-as-variation. That is, as "units" of analysis, rather than (or in spite of ) their contents, they serve as more than gestalts, but also as part of the infrastructure "gestalt perception." That is, they produce deutero-level effect, influencing how "the sequence of life experience, action, etc., is ... segmented or punctuated into subsequences or 'contexts' which may be equated or differentiated by the organism" (Bateson 1987, 213). But while Bateson thinks punctuation in terms of "contexts," I prefer to think of it in terms of "scales." That is, these disparate units of analysis (i.e. words, meanings, relations) have a contentless double at the deutero-level, that is nothing but a "unit." As an analogy, by weighing pounds of produce every day or yards of linen, you start to get a feel for those units, such that they begin to shape how you punctuate heft or length, more generally, and you can become quite adept at estimating the weight or length of unfamiliar objects. But the effect would only work at certain scales. That is, your skills of estimating pounds of a bag of produce wouldn't necessarily help you estimate the number of tons of coal on a freight line. In short, the units of analysis produce habits of perception that correlate with specific scales.
Analogously, in the realm of interpreting meaning, rather than estimating physical qualities, the units of analysis correlate with the scales of analysis, which will delimit the shape and character of the patterns you'll be able to find. For instance, despite what I might think about the quality or the usefulness of their analyses, the behaviorists found patterns in their data that really did exist in the data they collected. There certainly are going to be patterns of relations at the scale of sentences that correlate to beliefs. And these will enable you to objectify and produce characterizations of culture based on those patterns (whether the conclusions you draw from those patterns are warranted by those patterns is another matter). But that is just one scale at/with which social scientists (or interpreters of any kind) can punctuate their observations of (or engagements in) meaningful activity to produce and record data, and then to re-punctuate that data into cogent analysis. For instance, due to their divergent units of analysis, the symbolic anthropologists and post/structuralist anthropologists observe patterns and processes that emerge at their different scales. Even with their problems, each of these scales of analysis are capable of producing useful descriptions of sociocultural processes as well as insights into their operation and meaning, a fact to which the history ethnography testifies. What's more, the utility of punctuation is that it produces more than meaning, it produces the very possibility of perception, to the point that it becomes very hard to question. As Bateson puts it: "The practitioner of magic does not unlearn his magical view of events when the magic does not work. In fact, the propositions which govern punctuation have the general characteristic of being self-validating" (1987, 219). And again, "But a way of punctuating is not true or false. There is nothing contained in the propositions of this learning that can be tested against reality. It is like a picture seen in an inkblot; it has neither correctness nor incorrectness. It is only a way of seeing the inkblot" (1987, 219).
With its commitment to explanatory pluralism, PECE puts these different modes of punctuation on full display, inviting multiple interpretations of the same data in such a way as to produce traces of unique ways of punctuating data. And in that sense, it serves as an infrastructure for the kind of third-order learning that Bateson characterized as follows:
"The individual might learn to form more readily those habits the forming of which we call [deutero-learning].
He might learn to close for himself the 'loopholes' which would allow him to avoid [trito-learning].
He might learn to change the habits acquired by [deutero-learning].
He might learn that he is a creature which can and does unconsciously achieve [deutero-learning].
He might learn to limit or direct his [deutero-learning]."
