Notebooks

Analogy and Metaphor

16 Feb 2003 16:27

One of my pet peeves is people saying "metaphor" when they mean "analogy". A metaphor is a verbal construction, which can express an analogy. But an analogy can be expressed non-metaphorically, and metaphors can be so confused as to not express any coherent analogy (though perhaps then they express a bundle of conflicting analogies). To infer the analogy underlying somebody's thinking from the metaphors used in their speech is chancy; sometimes one set of metaphors is conventionalized for a given topic, but people can, on command, rapidly switch to a different set of metaphors to express the same ideas about the same topic. Yet it is generally the ideas we care about.

(Aside: in Flexible Bodies, a book about contemporary American ideas about immunity, the anthropologist Emily Martin recounts asking one of her immunologist informants to rephrase something without using an military metaphors. He was immediately able to do so, which suggests at the very least that, however his thoughts about the immune system were represented, the conventional metaphors of defense, infilitration, etc. were not essential. Martin, being a cultural anthropologist, completely ignored this point.)

Things get even more complicated when people build abstract models as, initially, representations of some real thing, and then proceed to analogize other objects to the model, using a metaphorical vocabulary drawn from the original object. Thus, to give an example close to home, certain parallel problem solving systems were first built as models of insect colonies, and so there is an entirely metaphorical vocabulary of "ants," "pheromones," "trails" and "foraging" associated with them. And there are now physical robotic systems which implement these computational models. But the analogy is really to the model, not to the nests, e.g., the models have nothing analogous to eggs or a queen, and so neither do the robots. If it turns out that we're stupendously wrong about how insect colonies work, the vocabulary will not go away or become inappropriate for the robots, since it's based on the analogy to the model, and the metaphors have become, to that extent, dead.

Understanding how analogy and metaphor work, and how they work together, is a fascinating topic that I am never going to contribute anything to.

permanent link for this note   RSS feed for this note

Notebooks: