Reading :: Models

Models: Representation and the Scientific Understanding
By Marx Warftofsky

I've been meaning to read this book forever. I think I first saw it cited in one of Yrjo Engestrom's early pieces (Learning by Expanding?) and was intrigued enough to see if the university library had it (it didn't). That was at Iowa State, and the same thing happened at Texas Tech and again at UT. Finally, after seeing it quoted again (this time by David Guile), I ordered the book.

The part of the book being quoted by Engestrom, Guile, and others is the chapter "Perception, Representation, and Action," which provides a taxonomy of types of artifacts. We'll get to it in a minute, but first, let's set up the book. It's in the Boston Studies in the Philosophy of Science, and it's composed of philosophical essays that Wartofsky published between 1953 and 1978. ("Perception, Representation, and Action" comes to us from 1973, the same year that Led Zeppelin released Houses of the Holy.)

In his introduction, Wartofsky explains that the pieces were written separately, but pulled together as a result of a series of four invited lectures he gave at MIT, in which he argued that "human knowledge itself has a history" — "that the nature of knowing, of cognitive acquisition itself, changes historically" (p.xviii). He traced these changes to human artifacts, including "tools and weapons" but also "language, forms of social organization and interaction, techniques of production, skills." And "the production of such artifacts for use ... was at the same time the production of representations" (p.xviii). These representations "are the precondition of so-called internal representations, that is, for the reflective activities of imagination, thought, conscious purpose" (p.xiv). Thus "human beings create the means of their own cognition" (p.xv). One can see why this line of argument would be interesting to Vygotskians such as Engestrom, Guile, and me.

Wartofsky further argues that "the cognitive artifacts we create are models: representations to ourselves of what we do, of what we want, and what we hope for. The model is not, therefore, simply a reflection or a copy of some state of affairs, but beyond this, a putative model of action, a representation of prospective practice, or of acquired modes of action" (p.xv). 

Thus we must consider two systematic issues. Analytically, what is the nature and function of a representation? Genetically, how did we generate the activity of representing itself, i.e., creating cognitive artifacts in the first place? (p.xvi). In these essays, Wartofsky says, he barely touches on the genetic question (p.xvi). 

Wartofsky goes on to argue that not all representation is linguistic: "representation is already involved in the very character of non-linguistic artifacts, insofar as they are themselves symbols, or are parts of systems of symbols, in which the artifacts themselves present meanings, intentions, relations, and come to represent to us the modes of practice involved in their production and use" (p.xvii). And "models are proffered truths. To proffer truth is the human means of acquiring knowledge. In this sense, cognitive acquisition, human learning is essentially mediated by representation. It is what makes theory possible" (p.xviii). Wartofsky adds this conclusion: "Strictly speaking, one may say that there is no human knowledge without representation; or more radically still, that there is no knowledge without representation" (p.xviii). 

With this argument in mind, let's turn to the aforementioned chapter "Perception, Representation, and Action," which most directly addresses this line of argument.

In this chapter, Wartofsky argues that "We do not perceive, and then act; perception is itself one of the instrumentalities or modes of action" (p.195). Thinking in these terms, we can understand historical human praxis as 

the fundamental activity of producing and reproducing the conditions of species existence, or survival. What is distinctly human about this activity ... is that human beings do this by means of the creation of artifacts. Their production, as distinct from the foraging, scavenging or hunting activity of other animals, proceeds by a transformation of part of the environment into an extension of the animal organs— as, e.g., tools are. But, in more generic terms, the 'tool' may be any artifact created for the purpose of successful production and reproduction of the means of existence. Therefore, the use of language for communication in this enterprise makes language itself such an artifact, or 'tool'; so too is the mode of social organisation, or of division of labor which is instrumental in the successful satisfaction of existence needs, or of the needs to reproduce the existence of the species. Extending the notion of 'artifact' or 'tool' still further, the acquisition of skills, in the processes of production (even at the level of foraging, scavenging or hunting, and prior to the introduction of agriculture or the domestication of animals) creates such skills as themselves 'artifacts', even where these skills do not entail the use of tools in the ordinary sense, but only the mastery of natural organs of the body, and of perceptual skills in pattern-or-cue-recognition, for the purposes of satisfying productive or reproductive needs. The crucial character of the human artifact is that its production, its use, and the attainment of skill in these, can be transmitted, and thus preserved within a social group, and through time, from one generation to the next. The symbolic communication of such skills in the production, reproduction and use of artifacts— i.e. the teaching or transmission of such skills is the context in which mimicry or the imitation of an action becomes a characteristic human mode of activity. It is, in effect, this ability to represent an action by symbolic means which generates a distinctive class of artifacts, which we may call representations. (pp.200-201, his emphasis)

Here, Wartofsky runs in parallel with Vygotskian ideas of mediation, and also emphasizes the cultural reproduction in which artifacts are enacted. He goes on to offer a typology of artifacts:

Primary artifacts: "those directly used" in "the production of the means of existence and in the reproduction of the species" (p.202), such as "axes, clubs, needles, bowls, etc." (p.201).

Secondary artifacts: "those used in the preservation and transmission of the acquired skills or modes of action or praxis by which the production is carried out. Secondary artifacts are therefore representations of such modes of action, and in this sense are mimetic, not simply of the objects of an environment which are of interest or use in this production, but of these objects as they are acted on, or upon the mode of operation or action involving such objects" (p.202). 

Through these representations, "nature becomes transformed, not only in the direct practical way of becoming cultivated, or shaped into objects of use ... it becomes transformed as an object or arena of action... nature itself has become historicized and socialized, and has come to be a representation of a certain mode of praxis or human action" (p.206).

Tertiary artifacts: Artifacts that are "abstracted from their direct representational function" (p.209); they "can come to constitute a relatively autonomous 'world', in which the rules, conventions and outcomes no longer appear directly practical," especially "when the conventions of representation ... become transparent" (p.208). He gives the example of a mimetic reenactment of a hunt or rehearsal for a hunt (p.207). These are "alternative imaginative perceptual modes, freed from the direct representation of ongoing forms of action, and relatively autonomous in this sense," but still they feed "back into actual praxis, as a representation of possibilities which go beyond present actualities" (p.209). While Wartofsky gives the example of reenacting a hunt, I think tertiary artifacts may cover examples such as formal logic that is abstracted from the actual problems it is trying to represent, or algorithms, or abstract formulas that can be applied to codified data (f=ma). 

Were these insights worth the price of admission? Maybe. The discussion above has been pretty well summarized by the people I've read who have drawn on Wartofsky's argument. But reading it myself, especially in close proximity to my recent readings on Boisot (who deals a lot with codification and abstraction) and the bio of Shannon (who was all about abstraction), has helped me to think a little differently about this typology. And what I think is that although this typology is a good starting place, providing a solid vocabulary for (for instance) critiquing Leontiev's treatment of tools, it is only a starting place for thinking through more complex issues such as how texts rerepresent our activities to ourselves. 

Still, it's important and challenging work. If you're interested in issues of artifacts and representations, check it out.