Symbols — from core symbols like the Virgin of Guadalupe to abstract ones like the whiteness of Melville’s whale — fix and generate ontological categories. How and why this happens is a question of deep interest to me, but that it is true seems obvious and well established. Human beings create symbols like plants produce oxygen, and symbol formation is inextricably bound to a defining trait of human beings — rich, discursive, and always already metacognitive language (human beings have always talked about talking, a practice that must be regarded as intrinsic to human language). Linguistics up to now, wedded as it has been to a Chomskyian Cartesianism, has missed this role, although philosophers have not lost sight of it. As Ricoeur wrote, “the symbol gives rise to thought.”
The relationship between language and symbolism is complex and (still) not well understood. My own view is close to that of the (admittedly discredited in its original form) generative semantics school, associated with George Lakoff. I believe that categories and rules are in some way generated by the transduction of meaning that takes place between neural representations of concrete objects. Discursive language — not “deep grammar” — tries to fix these meanings in propositional form, but the symbolic substrate has a dynamic quality, in no small measure do to its adaptive nature in response to what Merleu-Ponty called the “primacy of perception.”
With writing and then printing, and the monopolization of explicit knowledge, in the form of written records, reference works, etc., by governments, universities, etc., the relationship between discursive fixation and embodied symbols becomes tenuous and contested, resulting in a mind/body problem unfamiliar to ritual societies.
In any case, a number of practical observations follow from this tenet, which I will quickly enumerate, and hopefully take up later:
Human ontologies are not plans.
Human ontologies are overdetermined. That is, there is always more than one way to express an ontology The fixing of meanings will always fail if the goal is to create non-overlapping, non-redundant descriptions.
Human ontologies are rhizomic. In their natural form, ontologies are not hierarchical. Rather, the hierarchical representation is one form of serialization that works well because of its analogy to kinship (see Durkheim and Mauss, Primitive Classification).
Just as institutions require shared ontologies to function, so are institutions involved in creating the categories that nake up ontologies. Admittedly, assigning agency to institutions poses a number of questions that need to be answered; I won’t attempt that here. Essentially, I follow Mary Douglas (see video below), especially her How Institutions Think. The categories and rules that comprise human ontologies follow and enable a practical logic that in turn enables sustainable human interaction.
A corollary idea to this is that ontologies exist (in large part) to mediate social action. They are the result of human beings’ mutual calibration of individual cognition through collective interaction. Typcially, this calibration takes place through ritual. But media (old and new) — which grow out of but eventually displace ritual — also take on this role. (McLuhan’s frequent reference to ritual to describe the effects of new media is telling.)
“Social Life Makes the Categories”
The late Mary Douglas being interviewed in 2006
about her book Purity and Danger.
See the full interview and films at ScienceStage.
This is the first in a series of posts in which I define some of the tenets of comparative ontology, in order try to flesh out its significance to the work of making and using RDF vocabularies.
The overwhelming conclusion to be drawn from the ethnographic record is that human beings are surprisingly structured in their thinking and behavior, even when that behavior seems to be random and non-linear. Although it is a commonplace to observe that social life is inherently messy, unpredictable, and resistant to capture by physics-like laws (post-Einsteinian included) it remains true that patterns of culture are remarkably widespread and persistent. Languages, marriage practices, calendar systems, gift exchange systems, markets, etc. — essentially, any functional human institution — all rely on shared categories and rules to operate, and these are discoverable and describable. The mistake of the structuralists was to conceive of these categories and rules as logical in a formal, almost scholastic sense, like a plan that agents follow strictly. Instead, it is more likely that they exist as dispositions that constrain behavior and encourage improvization, as Bourdieu describes in his idea of the habitus (which he got from Mauss, by the way).
To the extent that formal RDF ontologies are meant to mediate human-computer interaction (and not simply allow computers to share information), ontologies should be designed to interdigitate with the categories of their human participants. Machine ontologies should be interoperable with human ontologies. They should be designed to encourage the symbiotic development and evolution of human collective representations (to use Durkheim’s expression), given the role of the networked computer and computer network as an institution in its own right.
Following up on my last two posts (here and here), I thought I’d flesh out a little more what I mean by Comparative Ontology — and, at this point, maybe I need a handle for the concept, like CO or “comp-ont.”
First thought: I imagine that a comp-ont vocabulary could be accomplished in two ways: (1) by just using existing vocabulary definition languages (RDFS and OWL) to create a new vocabulary, or (2) by creating a new vocabulary definition language using, say, Rodney Needham’s codification of English Structuralism in Symbolic Classification.
As an example of type 1, one might deploy the implicit ontology of Kautilya’s Arthashastra (with maybe a little of Machiavelli’s Prince thrown in) using OWL to create an “Enemy of an Enemy” (EOAE) vocabulary. This would be a useful vocabulary for describing political agents, linking them through the negative transitive logic of political alliance formation, with the result that you could actually predict or suggest alliances that don’t yet exist. I can imagine this vocabulary describing people (politicians, pundits, etc.), ideas, institutions, etc., and applied to daily news sources.
As an example of type 2, my question would be whether constructs like analogy or metonymy could be constructed on top of set theory (OWL) or if it would require a whole new system (and therefore logic of inference.)
Second thought: It strikes me that one way to bring home the value of comp-ont is to view it in the context of knowledge management (KM). Clearly there is a strong connection between KM and ontology (and the semantic web and linked data). But within KM, it is clear that ontologies can’t just be logical. Ontologies must be practical, and intelligible to the people in an organization who engage in ontology-mediated knowledge transfer. They must serve the purpose of mediating between individual and collective memory. But this is precisely what the great ontological systems from the ethnographic record provide — from Australian totemic systems to Mayan calendars. Viewed from the perspective of social memory, then, it makes a great deal of sense to build KM ontologies on general principles adduced from anthropological examples, and described in terms of structuralism.
So, I make the claim in my previous post on Comparative Ontology that comparativsim will help us make better ontologies and tools for using them. I realize I did not actually make that case. Let me try to do that now, although, to be honest, I view the claim as an interesting direction to follow (i.e. a hunch), not an established fact.
My argument is pretty simple. The Semantic Web, as everyone knonws, has a big problem — in order for it to succeed, producers of web content have to mark up texts using RDF, with its long URIs and arcane vocabularies. Most folks think this is a deal killer. In addition, many agree with Shirky’s claim that ontology is overrated in the first place. (The meme even makes an appearance in Michael Wesch’s viral “The Machine is Us/Using Us.”)
I hold out hope, though. I think tools can be created to make the data entry part a lot easier-withness Drupal 6. I also think content creators show a remarkable tendency to use tags, enter metadata, and engage in otherwise tedious SEO techniques if they know that they will get something out of it. This tendency is especially strong among digital humanitists, digital librarians, and other creators and curators of digital collections. (Think of the time people have put into TEI!) Right now, they don’t see a lot a ROI from using RDF. But that is changing, and faster than many think.
As is often remarked, it’s a classic chicken and egg argument. But such problems are actually liberating, since you can intervene at any point in the cycle and have an effect. One juncture in the cycle is in the design of vocabularies.
The problem now is that vocabularies and ontologies are just not sticky enough, and I think a big reason for this is that they are written from the perspective of symbolic logic and set theory. These frameworks, though comprehensive, are not human (enough). On the other hand, there is a whole tradition of describing cognitive systems called structuralism, which has the advantage of being empirically grounded in the longstanding comparative study of culture. It stands to reason that if we created vocabularies using more “human” constructs-such as analogy, metonymy, etc.-and vocabularies that recognized the role of verbs in predicates, then we might be able to produce more sticky ontologies.
It also opens a way to harvest the latent ethnographic data contained in folksonomies.
The Semantic Web has popularized the concept of ontology. The usual definition one sees for the term is Gruber’s “specification of a conceptualization” (Gruber 1993). The definition could use some unpacking. Essentially, it refers to the idea that the terms used by AI agents should be defined by reference to a shared schema which classifies those terms. The shared schema, or specification, simply defines how terms are related to each other in a graph. More than a simple data model, where the relationships between entities or objects are not explicitly defined, an ontology also names the possible relationships between terms. So, instead of just having “address” as a field of the entity “person,” an ontology will explicity assert that a person has an address. So verbs (which are often buried in “predicates”) are an important part of ontologies. In fact, one way to think of an ontology is to call it a data model with verbs. (Sort of.)
Ontologies are useful for programming agents, which are just programs that consume and produce information on their own. (I call them infophages.) So, when you program an agent to respond to a term, you have the agent’s program refer to an ontology to disambiguate the term among synonyms, and to make logical inferences about the term. For example, if the agent encounters the term “Socrates,” it will be found to be specified as a member of the class “Human” and since “Human” will have the property “is Mortal,” the agent can transfer the property to “Socrates” too. (Truth, or validity, is just traceability within a reference graph.)
Or, if the agent encounters the word “Madonna,” and that term is formally specified as part of an ontology (say, by using Tim Berners-Lee’s URI method), the agent will be able to trace its class to either “Pop Singers” or “Religious Figures” or “Christian Iconography” or whatever.
Sometimes-often-ontologies are described as if they provide “meaning” for agents. Even the term Semantic Web implies this. When introduced to novices, the Semantic Web is often introduced as bringing “meaning” to the web. I find this explanation misleading at best. Meaning is much to too important of a word to describe what is going on here, which is simply a layer of classification being added to what would other wise be a list of terms. Ontologies are just systems of classification for terms, a second-order set of terms that (1) increases the probability that an agent can process a term in a way that users will expect, and (2) adds a layer of connectivity, above the raw verbiage on the web, that decreses the average distance, or degress of separation, between any two terms.
So, ontologies have two properties that make them useful and, as I want to argue, usefully viewed from a comparative perspective. First, they introduce verbs into the mix of data modelling. Second, they are just systems of classification-like Australian totemic systems, Mayan calendars, and Western philosophical ontologies such as Aristotle’s and Leibniz’s.
Comparativism will help us create better ontologies, and better systems for using ontologies, in two ways.
First, comparativism can train our attention on the great rabbit warren of words and meanings that lies at the heart of the ostensibly neat world of triples-I refer to the predicates. Although the nouns that comprise the vocabularies for subjects and objects in RDF can be neatly specified as analytic taxonomies of terms, predicates are subject to no such rules, and can contain within themselves whole sentences. Unlike the nouns, predicates mask a great deal of what Kant would call “synthetic judgements.” I propose a sociolinguistic approach to the use of verbs embedded in the predicate systems of linked data vocabularies that will provide a better basis for crafting predicates.
Second, comparativsm can help us move beyond set theory-useful as it is-to consider other totalizing schemes for organizing ideas, such as those described by anthropologists. I mentioned totemic systems, calendars, and Western ontologies. But there are many others to consider. What is more, anthropologists have done a great deal to describe, classify, interpret, and even explain these. Such an approach would be grounded in the following works, all of which define or grow out of the structuralist tradition within anthropology:
Emile Durkheim and Marcel Mauss, 1903, Primitive Classification.
Louis Dumont, 1966, Homo Hierarchicus.
Michel Foucault, 1966, The Order of Things.
Claude Lévi-Strauss, 1968, The Savage Mind.
Victor Tutner, 1969, The Ritual Process: Structure and Anti-Structure.
Mary Douglas, 1970, Natural Symbols.
Pierre Bourdieu, 1977 [1972], Outline of a Theory of Practice.
Alfred Gell, 1975, Metamorphosis of the Cassowaries.
Rodney Needham, 1975, Polythetic Classification.
Edmund Leach, 1976, Culture and Communication.
Marshall Sahlins, 1978, Culture and Practical Reason.
Diana Forsythe, 1993, “The Construction of Work in Artificial Intelligence.” *
Steven Bird and Mark Liberman, 1999, “Annotation graphs as a framework for multidimensional linguistic data analysis.”
Lev Manovich, 2001, “Database as Symbolic Form.”
John Unsworth, 2001, “Knowledge Representation in Humanities Computing.”
Stephen Ramsey, 2005, “In Praise of Pattern.”
Tim Berners-Lee, 2006, “Linked Data-Design Issues.”
* These were added after the original post. I reserve the write to amend and annotate this list at any time …
Here are a few qualifiers.
Number one, I don’t think there are, strictly speaking, any foundational texts in the digital humanities. Not in the way that physics can claim Newton’s Principia or biology Darwin’s Origin. I am not sure if this situation is due to the (perennial) infancy of the field-if, indeed, it is a field. (I think of it more as a cross-displinary methodology.) Moreover, my saying so certainly isn’t due to any distaste on my part for the concept of a canon. Instead, there are, roughly corresponding to the pre- and post-war era of the previous century, a number of loosely related essays that adumbrate a set of ideas which subsequent generations of people who call themselves digital humanists have been unpacking. The latter have produced a number of essays which the majoritoy of digital humanists will have read, and these may be called foundational, in the sense of a shared discourse. Fair enough. But none of these texts can claim the status of having defined a method or a domain that we can, in retrospect, claim as distinctly concerning the digital humanities. Also-and here I will be controversial-I believe that this particular corpus has had the effect of producing, though the hyercoherence that can affect small “thought collectives” (Fleck), a rather narrow set of concerns which have put the field into a groove from which it would do well to extricate itself.
Number two, since I consider the digital humanities to be at once critical and practical, these texts come from both angles, one set concerned with method, the other with historical context.
Number three, parts of this list are, as you’ll quickly see, pretty specific to me, and my background as an American cultural anthropologist with strong English and French influences. These are texts that have been foundational to my conception of the digital humanities, and which have made a difference to my way of thinking about textuality, digital textuality, and what happens to text when it becomes digital. But it is not entirely idiosyncratic. In its defense, I would argue that the digital humanities is more closely tied with the structuralism of the 1950s and 60s than is usually recognized-and indeed, more than structuralists are prepared to admit. (There is an important chapter of intellectual history that needs to be written here, concerning the close but relatively hidden relationship between structuralism and the “cybernetic moment.”) Moreover, the anthropological angle is always worth pushing, insofar as the discipline, before its reflexive self-implosion in the 1980s and 90s, bequeathed the culture concept on which the realignment of the humanities and itnerpretive social sciences has been constructed. From an anthropologist’s view, the wild success of cultural studies and cultural history has the sad sweetness, the tristes tropiques, that attends the simultaneous death and birth of cultures.
I call this blog the “transducer” in honor of an idea that I want to cultivate and expand into a framework for a philosophical anthropology I have been gestating since my undergraduate days, when I spent most of my cycles trying to process Christianity, capitalism, Darwinism, Marxism, and physics. I now believe the concept of transduction is the, or a, key to understanding the relationship between meaning and causality, mimesis and mutation, the fundamental ontological dualism that philosophical monism can never seem to dispense with except by periodic calls to burn the books of theologians and metaphysicians. (This is an enlightened tradition that includes Hume, Ayers, and now any number of New Humanists who have no time for this sort of nonsense.)
The theory of transduction, as I want to develop it, is a possibility opened up by Shannon’s theory of information (so-called). In focusing on this concept, I am inspired by the at this point undervalued work of Roy Rappaport and by the brilliantly muddled works of Gregory Bateson, to whom everything interesing I read nowadays concerning teaching, learning and technology is more a less a footnote.
Transduction is the transfer of information that attends every transfer of energy. For example, in any mechanical interaction, energy is transferred from one body to another. But information also passes: the amount of energy transferred, the direction of movement, the frequency of the event if it is repeated, the time of the event, etc. Measurable indicators such as velocity and position are part of the information that inheres in the event.
Transduction is most apparent — and useful — when energy is converted from one form to another. A well-known example is the transduction that occurs when mechanical energy is converted into electrical energy through a phonographic needle designed to optimize the attendant transfer of information. The needle is part of a larger system of transduction: sound waves are transduced by a microphone, which are transduced into a recording medium, which is then transduced back into electricity, which transduces the sound through the mechanical vibration of a speaker.
The theory of transduction follows from the hypothesis that everything (and everyone) acts as a needle in a vast economy or ecology of signal transmission. We are all transducers made of countless transducers, and we are probably elements of larger transductivities, such as the economic system and so forth. And we are constantly transmitting and transducing signals which “contain” messages. Here it becomes interesting — where do the messages come from? Do we create messages? If so, how? Are there really messages, or is it signals (traces, différance) all the way down?
Pop quiz: How many needles dance in the head of an angel? (Angels, of course, are God’s messengers, Judeo-Christian versions of Hermes and Mercury. McLuhan had something to say about angels as well.)
The term transduction, of coruse, has been appropriated by genetics, where it refers to a process in which DNA is transferred from one bacterium to another by a virus, or where foreign DNA is introduced into another cell through a virus. This too is transduction in the sense that I mean it — that is, it’s not just a semantic appropriation. But it introduces the notion of mediated transduction by means of an agent — in this case, a virus. In this case, the virus plays the role of the wax record (to keep the analogy simple) in the sequence of transductions I described above.
What is interesting about viral transduction is that it provides an example of how simple, unmediated transduction produces complex, mediated transduction. If, as I posit above, the universe is a vast system of transduction, then, apparently, elements of that system evolve and emerge to play roles in that system. The virus seems a clear case of this, evolving only to perform that transductive work of a cell-based, biological system, which itself emerged through transductive reproduction.
And this is another concept that seems describable in terms of transduction — reproduction. For what is transduction but the reproduction of information? But always an incomplete reproduction, as not all information passes through. As Bateson famously wrote, information is any difference that makes a difference. That is, only information that survives transduction is information. So, transduction is a kind of reproduction, but it is also always a misprision, a creative misreading.
An interesting consequence of the theory of transduction is that you can’t measure transduction without introducing more of it, and therefore becoming subject to Bateson’s law (you only get differences that make a difference). This fact normally doesn’t bother us, but it really comes out in relativity theory, where Einstein shows that time is, in a very real sense, the meausrement of time.
Other examples of transduction include walking and leaving footprints, the impression on a wooden surface left by thrown rock, the decaying of carbon 14, settlement patterns left by the inhabitants of population centers, the Van Eck “phreaking” described in Stephenson’s Cryptonomicon, biological evolution of genotypes, learning, etc. Yes, evolution and learning, what Bateson wrote about mostly. The concept is recursive in the sense that transductive elements can form transductive systems which are in turn transductive.
Ultimately, if all this is true or coherent, we, as language-using humans, are vastly complex transducers of signals that go way, way, back, picking up cosmic, genetic, cultural, and social signals as we go through life, as wayfarers (to use Walker Percy’s use of Augustine’s image).
And this is how I view blogging . The Internet is probably the first human institution (and it is an institution, more than it is a “technology,” whatever that means) built explicitly on the principles of transduction. We are each transducers of the messages that pass through our “post,” to use Lyotard’s image. It’s a good image, one that recalls Pynchon’s image of the Pony Express in The Crying of Lot 49 and Melville’s image of mail in Bartleby. For, from the perspective of transduction, what worse hell can be imagined than the dead letter office?
