Ostension, abstraction and ambiguity

I wanted to continue with the topic of morals, this week, but the press of work in my “day job” didn’t leave me time to think through some things I wanted to talk about. So I’m posting this piece, which I wrote spontaneously in my journal a couple of days ago. The only connection is that I was reading the De Waal book, Good Natured, that I mentioned last week, and I was thinking about the evolutionary origins of human cognition. Anyway, hopefully I’ll get back to morals, next week. By then I should have finished De Waal, at least.

Bertrand Russell, among others, has pointed out the limits to definition. Words being defined in terms of other words, eventually one reaches the point where further verbal definition is possible only by permitting circularity. In formal languages, such as mathematical theories, the solution is to leave some terms undefined in the subject language, relying instead on definitions in some background language or “meta-language”.

In natural languages, the equivalent of meta-language definition is definition by ostension. One points to an object and says, “This is a table.” Or (to use W. V. Quine’s favorite example), one points to a furry animal and says, “Rabbit.”

What Russell doesn’t really point out, at least in my readings thus far, is the degree to which definition by ostension involves a process of abstraction. Quine makes much of the ambiguities involved – but this is a different, although related, point.

To identify a table, you need to determine the boundaries of “table”. You need to determine which particular parts of experience you are isolating (mentally) to define them as “table”. Your audience needs to perceive things similarly for communication by ostension to be meaningful. Especially to identify “table” as a general term, or identify a class “table”, you both need a similar functional/pragmatic relationship to a table object, otherwise the general term makes no sense. This is different than distinguishing between a rabbit and an undivided collection of rabbit parts (to consider Quine’s example). This is distinguishing between the rabbit and the ground it hops around on. An intelligent ant might find it impossible to distinguish between ground and rabbits by ostension, and might, in fact, find each term to be an incomprehensible generalization, rather like a human might react to a class containing jelly beans and squid.

The point is that the simple act of ostension, the most basic form of definition, involves a process of abstraction, and is therefore partly a function of the cognitive apparatus of the communicants. The cognitive apparatus in turn evolved into what it is as a function of the way the thinking organism interacts with its environment. This means, of course, that the cognitive component to ostensive definition, while it is “subjective”, is in no sense arbitrary – except, perhaps, at the margins of utility. It also means that since conspecifics share so much of the ways in which they functionally interact with their environment, members of the same species can generally go quite far in communicating meaning by ostension. Quinean ambiguities can be seen in context, here: there really is little functional difference, to a human, between a rabbit and an unseparated, reasonably complete, set of rabbit parts – hence the translational ambiguity that Quine finds.

References: Quine used his rabbit example frequently, most thoroughly, if I remember correctly, in Word and Object, but also in the essay “Ontological Relativity”. I’m damned if I can remember where I read Russell on definitions. (It was longer ago.) Quine, by the way, must be rolling over in his grave at my loose correlation of the notions of “class” and “general term”.