The Way We Think: Conceptual Blending and the Mind's Hidden Complexities. Gilles Fauconnier and Mark Turner. xx + 440 pp. Basic Books, 2002. $35.
Let's suppose that you're looking for your cat, and you hear someone say, "The cat is on the mat." You immediately understand—that is, you put together your ideas of the searched-for cat, the remembered mat and the relation of being on something, and you have the thought of the cat being on the mat. You are using (albeit on an unconscious level) the tools of traditional empiricism and logic—abstraction (deriving the concept "on" from your familiarity with lots of "A is on B" situations) and instantiation (recognizing the sought cat and remembered mat as a concrete instance of the abstraction). This is a case of conceptual blending: You have taken thoughts or ideas picked up in one situation and put them to use in another.
Now suppose that you've just walked into the philosophy department of a university, where a vigorous debate is going on between the local Kant and Hume enthusiasts. "The Kantian is on the mat," someone whispers to you. In this case, understanding what has been said is a bit trickier, since the philosopher isn't literally occupying a position on a physical mat. To get the intended message, that the Kantian is losing as badly as a wrestler who has been thrown to the mat, you must grasp what philosophical arguments have in common with wrestling matches—that they are forms of competition, with winners and losers. And you must recognize that features of one are being borrowed to describe the other—that wrestling vocabulary is being used here to convey vividly and efficiently who is losing the debate and how badly. Here again you are engaged in conceptual blending—using ideas picked up in one framework to think about another.
For Bertrand Russell, Gottlob Frege and their successors in the flowering of logic in the 20th century, it was important to understand what is going on in descriptive, factual statements like "the cat is on the mat." For the purposes of logic and science, analysis of statements like "the Kantian is on the mat" seemed to them far less central.
Gilles Fauconnier and Mark Turner disagree. Cognitive scientists at the University of California, San Diego, and the University of Maryland, respectively, Fauconnier and Turner are part of what seems to be a paradigm shift in cognitive science, from the study of the unambiguous formulae of first-order logic to analysis of tropes (figures of speech, once studied only in literature departments). In The Way We Think, they argue that analyzing tropes and other imaginative, concept-stretching and mind-bending uses of language reveals more about the dynamics of conceptual blending, and that conceptual blending is at the heart of how we think.
Much of the discussion hinges on what the authors call the network model of conceptual integration. A network involves several mental spaces, or small conceptual packets, which can be occupied by disparate and partial memories, perceptions, scenarios and the like, with elements and aspects mapped onto one another.
Fauconnier and Turner identify several varieties of conceptual networks. A simplex network takes values from one place and plugs them into a framework. For example, the cat and the mat might occupy one mental "input" space while the familiar relation or framework of A being on B occupies another; when the cat and the mat are plugged into the framework, an "output" space, or blended space, is then occupied by the thought that "the cat is on the mat."
A mirror network is one in which all of the spaces share an organizing frame that specifies the nature of the relevant activity, events and participants; the input spaces mirror each other in having this same frame. The authors give as an example a newspaper illustration depicting six runners on a quarter-mile track, each of whom has at some point set a world record for running a mile; they are positioned on the track according to their record-breaking finish times, with the latest record-holder crossing the finish line and those with slower times at corresponding distances behind him (their positions calculated on the assumption that each race was run at a uniform speed). The organizing frame is "running the mile and breaking the record." As this example shows, a mirror network makes it possible to compress time.
A single-scope network uses ideas from one mental space (for example, A and B are wrestling, and A throws B to the mat) to enrich the description of an event from a framework with a certain generic similarity (C and D are arguing about Kant and Hume, and C is losing badly); the blend is "the Kantian is on the mat."
At the heart of the book is the double-scope network, which combines two possibly conflicting frameworks and comes up with something new. A favorite example is the computer desktop, which combines the framework of offices and the framework of computer operations to give us something novel—and quite wonderful. According to the authors, double-scope thought is the mental capacity that drove the development, during the late Paleolithic period, of "art, science, religion, culture, sophisticated tools and language." Double-scope blending isn't simply central to their theory; for them, it's central to the human condition. The authors may be quite right that creative borrowing from old ways of thinking to help structure thought about something new put us on the path to human language and culture. It's as appealing as any theory along those lines that I know of.
Still, it will be hard for them to use double-scope blending to explain how mind and language evolved in the world unless they bring the world as well as the mind into their theory a bit more robustly. In a case that they use as an example of a simplex network, "Paul is the father of Sally," one space in the network is occupied by the frame of human kinship known as the family and another space is occupied by the individuals Paul and Sally. Conceiving of Paul as the father of Sally blends those two spaces. Fauconnier and Turner comment as follows:
The blended space in such a network is compositional in the sense that the entirety of the relevant information from both inputs is brought into the blend. This composition is truth-conditional in the following sense: The sentence counts as "true" in a world if the blend fits the current state of that "world" (i.e., if Paul is indeed the father of Sally).
The quotation marks around "true" and "world" are ominous. Are the authors suspicious about truth and the world?
To return to my single-scoped example, I won't understand the comment that "the Kantian is on the mat" unless I realize that the speaker is telling me something about the philosophical argument—something about which he is either right or wrong.
Truth is a concept that may seem out of place with the imaginative blends of double-scoped thoughts. But surely what was important in evolution wasn't just having all of these great double-scoped thoughts. It was that they proved efficient as a way of communicating and useful as a way of thinking. Double-scoped networks borrow frameworks from phenomena of one kind and use them to organize phenomena of a different sort. Such networks can make truths about the unfamiliar domain "pop out." The desktop metaphor worked to enable a generation to think about computer operations not only because desktops were familiar to many for whom computers were a mystery but also because for lots of purposes the familiar desktop relationships adequately represent the unfamiliar computational functions.
I think that in hooking together their theory of the structure of thought and imagination, the authors have underestimated the usefulness of the structures of modern logic and formal semantics. There are now logics of diagrams, fuzzy-logic programs (which run everything from washing machines to elevators) and lots of other very distant cousins of the sentences of first-order logic. We cognitive scientists need double-scoped thoughts to understand how conceptual blends are useful in dealing with information about the world; we need some formal semantics with our psychology, and the conflict and creativity its inclusion would spark.
There is much to recommend the authors' input into such a blend. The writing in The Way We Think is lively, and the examples are plentiful, each usefully given a name so that the authors can easily refer back to it. The analyses of language are insightful and mostly convincing, the diagrams are helpful, and the jokes are entertaining. Any student of language and thought will learn a great deal from this fascinating book.—John Perry, Philosophy, Stanford University