Yesterday I published Redefining the Coming Singularity – It’s not what you think, in which I speculated about the future of computing technology. Those speculations, however, were grounded in observations about cultural change in the past. In particular, I argued that there are 'singularities' in which an old 'ontology' gives way to a new one. This post from 2011 is about that sense of ontology, ontology as the basic framework of a conceptual system. My principle example is "salt" vs. "sodium chloride," which designate more or less the same substance, but a terms in different conceptual systems, different ontologies. Ontology in Knowledge Representation explains this more fully.I have a long-standing interest in ontological thought, how we think about ontological matters: animal, vegetable, or mineral? As such I’m not directly interested in what really exists in the world, about the ‘ultimate constituents of reality’ – though my vague understanding is that object-oriented ontology (OOO) questions the ultimacy of some constituents over others – but simply how we think about things. That interest is one aspect of my general interest in the cognitive sciences and, in particular, knowledge representation.
Knowledge representation originates in the problem of programming computers to ‘reason’ about the world in a flexible way. It’s one thing to perform complex calculations given numerical data and the appropriate procedures. It’s a somewhat different thing to keep tabs on things and stuff, a parts inventory, the population of a country by region, state, county, and city, town, or village. But to reason about, say, medical diagnosis, or simply whether to take an umbrella with you when you go out for a walk, that’s more subtle. How do we model the knowledge required?
Those sort of questions are the background of my observations below. The point is simply that we’re talking about very explicit models of thought, models that can be programmed into computers. That’s a bit different from simply thinking about thinking, even in a reflective way. No you’ve got to get a machine to do a convincing job of mimicking thought.
Let’s start with a consideration of salt and sodium chloride, which I’ve excerpted and revised from an unpublished article, “Ontology in Knowledge Representation.” When we’re through that I’ll consider some implications of living with mutlple ontologies.
Of Salt and Sodium Chloride
We all know that salt and sodium chloride are, physically, pretty much the same. Conceptually they are very different. Salt is a white granular substance with a certain taste, and that taste is more salient in our understanding of salt than its appearance or texture. After all, the taste tells us of salt's presence even where there is no white granular substance to be seen or touched. Salt is thus rather adequately defined in terms of sensory perceptions.
Sodium chloride is a chemical compound whose molecules consist of one atom of sodium and one atom of chlorine. What in that definition is a sensory perception? "Compound," "atom," "molecule," these are all abstract. And if we start looking behind these abstractions one route will lead us to meter readings on laboratory instruments while another route will lead us to hadrons, leptons, and a handful of forces, strong, weak, electromagnetic, and gravitational, not to mention itty-bitty ‘strings’ in 9, 10, 11 dimensions. The conceptual domain in which we find "sodium chloride" is thus quite different from the one in which we find "salt." To borrow from the language of set theory, the extensions of those terms are much the same (note that salt ordinarily contains impurities which aren't in the extension of "sodium chloride") but their intensions are different.
"Salt" is defined in the domain of every-day commonsense knowledge while "sodium chloride" is defined in the specialized domain of chemistry. These domains have different ontologies. Philosophers and physicists do worry a great deal about ontological questions - what is the world ultimately made of? Except to the extent that the mind/body problem is properly the concern of artificial intelligence or cognitive science, questions of ontology are not central to us. But, the representation of knowledge is central, and ontological knowledge is clearly important in thought.
Animals know salt, at least by taste. The substance itself can vary widely in appearance and naturally occurs in mixtures with other substances rather than as more or less white grains. Though I have no citation readily at hand, I believe that saltiness is a primitive element of the taste system. Which implies that we are born with a taste for salt. It’s in our blood – both literally and figuratively.
I don’t know how old children are before they can think of salt as a substance different from other substances, with its particular diagnostic features. I do know that there was a time in my life when I couldn’t detect any visual difference between table sugar and table salt. They’re both white granular substances. However, when you cover a piece of buttered bread with table salt – which is what I did – you don’t get a sweet tasty treat – which is what I was expecting when I poured that white granular substance on my bread. You get the overwhelming taste of salt salt salt: not what I was trying to achieve, not at all.
It wasn’t until middle school or perhaps even later that I learned about sodium chloride, NaCl, and the conceptual system in which it is embedded. That came as a great revelation, that ordinary salt could also be this exotic sodium chloride stuff.
This personal progression has its historical correlate, indeed the personal presupposed the historical. For most of human history, while everyone knew about salt, no one knew of sodium chloride. The notion of sodium chloride is embedded in an atomic theory of matter that was worked out in the European West during the 17th, 18th, and 19th centuries. And during the 19th century those atoms came to have subatomic particles, which then proliferated during the 20th century. In this, the 21st century, our ability to theorize about the subatomic domain exceeds our capacity to observe it. So we’re in something of a mess, pursuing the ultimate constituents of matter.
But that’s a side issue. The central point is simply that we live within multiple conceptual ontologies. Everyone, man, woman, child, and beast, knows about salt. Only those with a certain type of education know about NaCl, but it’s they salt conceive of at the dinner table, not NaCl. And among those, only a very few really grasp the intricacies of the subatomic world. We negotiate our world with multiple ontologies in our heads.
Thomas Kuhn called them paradigms (well, paradigms are more than ontologies, but they include ontologies), and they seemed exotic when he did so. But they’re not exotic at all. They’re the stuff of widespread modes of thought.
And so it is with the world of biology as well. Work on folk taxonomies in many cultures makes it clear that all peoples recognize such classes and distinctions as animals vs. plants; fish, birds, four-legged beasts; trees, grasses, and bushes. It takes a more sophisticated conceptual system to classify whales and, shall we say, cattle together as mammals. And it took the invention of the microscope to observe single-celled life forms of all sorts.
What of all the machines and contraptions of the Industrial Age and beyond. I’ll bet the people who work with them do so through specialized ontological concepts. Ontological multiplicity is a feature of human thought.
But What’s Really Real?
That question, as I said at the outset, is off the table. The study of ontological reasoning won’t tell you that. That study tells us how conceptual ontologies guide our thought. It doesn’t tell us what to think. It tells us about category mistakes – green ideas, trustworthy mush, and the like – but not how to conceive new categories, and hence to think new kinds of thoughts.
Still, it is a useful thing to know, that ontological reasoning is an aspect of ordinary thought. And especially, that we negotiate our lives in multiple ontologies. But whether or not there is some ‘meta’ ontology that encompasses them all, well, I doubt it.