Nirad Das (an Indian painter): You wish me to be less Indian?
Flora Crewe (an English poet): I did say that but I think what I meant was for you to be more Indian, or at any rate Indian, not Englished-up and all over me like a labrador and knocking things off tables with your tail.… Actually, I do know what I mean, I want you to be with me as you would be if I were Indian.
Das: An Indian Miss Crewe! Oh dear, that is a mental construction which has no counterpart in the material world.
Flora: So is a unicorn, but you can imagine it.
Das. You can imagine it but you cannot mount it.
Flora: Imagining it was all I was asking in my case.
Das (terribly discomfited): Oh, Oh, my gracious! I had no intention—I assure you …
Tom Stoppard, Indian Ink
PARABLE DISTRIBUTES MEANING across at least two stories. But nowwe have seen that a third story—a blended story—typically plays a role. We have also seen that a blend can have more than two inputs, and that blending can happen recursively—a blend can be blended with other inputs to create yet another blend.
In fact, parable typically distributes meaning over many spaces. The aggregate meaning resides in no one of them, but rather in the array of spaces and in their connections. We know each of the spaces, and how it relates to the others, and what each is good for. None of them replaces another. Meanings, in this way, are not mental objects bounded in conceptual places but rather complex operations of projecting, blending, and integrating over multiple spaces.
We have often seen that an element in one input story can have a counterpart in the other. The donkey and Shahrazad are counterparts. So are Great America II and Northern Light, the two Buddhist monks, the two paths they travel, the rain and the message, and the reaper and Death. It is not possible to blend two stories without some counterpart connections between them to guide the blending.
Input stories have counterparts because they share abstract structure. Consider the boat race. There is a frame for a voyage by boat. It contains a boat, a path, a departure point, a destination, and so on. Each of the input stories for the boat-race blend has this frame structure, which provides counterpart relations between them. Different kinds of abstract structure can be shared by two stories: category structure, frame structure, role structure, image-schematic structure, and so on. Sometimes, two inputs will share abstract structure because a conventional metaphor has established that shared structure. Sometimes, two inputs will share more than one kind of abstract structure. For example, the inputs for the riddle of the Buddhist monk share frame structure (a person walking along a path) just as the inputs for the boat-race share frame structure (a boat voyage), but there is also identity structure shared in both cases: The monk in one space counts as “identical” to the monk in the other space, and San Francisco in one space counts as “identical” to San Francisco in the other space.
Now consider the expression “If I were you, I would have done it,” said by a man to a woman who declined earlier to become pregnant. The two input spaces share structure: In each there is a human being, with human intentionality, possibility, and standard relations between intentionality and action. This shared structure gives counterparts: The man and the woman are counterpart human beings, even though they have different identities; their preferences are counterparts, even though those preferences are not identical; their actions are counterparts, even though their actions are not identical. In short, there is a frame structure shared by the two input spaces, and it provides counterpart relations between them.
I will say that the abstract structure shared by input spaces resides in a generic space. The generic space indicates the counterpart connections between the input spaces.
The central question is, is this generic space just a name for structure shared by the input spaces, or does it have an actual conceptual existence of its own? Can this abstract structure be manipulated and used in a way that does not entail manipulating the input space in which it resides? George Lakoff and I have given one argument for the existence of generic spaces, as follows.
When we read a proverb in a book of proverbs, and so have no reason to connect the meaning of the proverb to any specific target, we arrive at a generic reading. For example, “Look before you leap,” found in a book of proverbs (or in a fortune cookie), will be interpreted generically. Similarly, “When the cat’s away, the mice will play” presents a source story of mice who behave in a restricted fashion when the cat is around but who behave with fewer restrictions when the cat is gone. The generic-level information in this story can be projected to a generic space with an abstract story: One agent or group of agents constrains another agent or group of agents, and when the governing agent is inattentive, the otherwise constrained agent or agents behave more freely. We can reach this generic interpretation even if we have no specific target onto which we wish to project it. So the generic space has a conceptual existence.
Lakoff and I have called this kind of projection from a source story to a generic story GENERIC IS SPECIFIC: generic information, often image-schematic, is projected from a specific space to give structure to a generic space. Of course, this generic space “applies” to the specific space from which it came. Once the generic space is established, we may project it onto a range of specific target spaces. The generic space constructed out of “When the cat’s away, the mice will play” can be projected onto stories of the office, the classroom, infidelity, congressional oversight committees, computer antivirus utilities, and so on, over an unlimited range.
Blends can be constructed if two stories can be construed as sharing abstract structure. The abstract structure they share is contained in the generic space that connects them. Consider, for example, the riddle of the Buddhist monk. Its generic space has a single journeyer taking a single journey from dawn to sunset, over a single distance along a single path. The generic space does not specify the direction of the journey (up or down), the date of the journey, or the internal form of the journey (starting and stopping, moving slower or faster). This degree of inspecificity allows the generic structure to be projected equally well onto the space of the ascent on the first day, the space of the descent on the second day, and the blended space where both journeys occur on an unspecified day. The riddle of the Buddhist monk involves two input spaces, a generic space, and a blended space.
It often takes work to find a generic space that fits two input spaces; there are often alternative generic spaces that might connect two input spaces. But in some cases, a generic space has been constructed repeatedly between two spaces. It has come to structure the two input spaces and to establish what seem to be fixed counterparts. The generic space has become fully available from each of the input spaces. In that case, the generic space becomes invisible to us. If the blend between the two spaces is conventional or minimal, the blend may also be invisible to us. In that situation, we are likely to detect only two input spaces, without a generic space or a blended space. We are also likely to detect what seems to be a direct projection between them, involving a set of fixed counterparts. This situation arises in basic metaphors like LIFE IS A JOURNEY. In basic metaphors, it looks (at first) as if the projection goes directly from one space (the source) to another (the target), without involving a generic space or a blended space. The projection seems to be one-way and entirely positive. It seems to have fixed parts: For example, the traveler projects to the person living the life; the beginning of the journey projects to birth; the end of the journey projects to death; the distance traveled projects to the amount of time lived; obstacles project to difficulties; guides project to counselors; fellow-travelers project to people with whom life is shared; and so on. These are the projections of basic metaphor analyzed insightfully by George Lakoff and Mark Johnson in Metaphors We Live By. The invisibility of the generic space and the blended space (except, of course, on analysis) produces a minimal phenomenon that looks like direct projection from one conceptual domain to another.
When a deeply entrenched projection is expressed in deeply entrenched vocabulary, especially vocabulary so entrenched as to seem to be the natural way to discuss the target, then we do not notice the projection consciously and have no need to notice the generic space or the blended space. For example, the phrase “intellectual progress” (from THE MIND IS A BODY MOVING IN SPACE) takes its noun from the source and its adjective from the target and can be thought of as evoking a blend, the way “fossil poetry” evokes a blend. But “intellectual progress” seems normal and “fossil poetry” seems imaginative. For “intellectual progress,” the underlying generic space is absolutely established; the vocabulary of the source (JOURNEY) has been effectively shifted to this generic space during the previous history of the language and is available to all speakers now to be applied linguistically wherever the generic space is applied conceptually; the conception of the target (MIND) by projection from this generic space is very deeply entrenched; and the phrase “intellectual progress” is a conventional phrase for the target conceived in this fashion. It is accordingly difficult in this case even to notice the generic space.
By contrast, the expression “mental journey,” which shares its linguistic construction (noun from source, adjective from target) and its conceptual projection with “intellectual progress,” is less conventional as language, and so in this case it maybe easier to see the projection and to recognize the generic space. “Ethnic cleansing,” yet another example of the same linguistic construction, is not at all conventional as an expression. True, the generic projection of the scenario of cleansing is highly conventional; the resulting generic space has been applied to many specific targets and the source vocabulary has been thoroughly shifted from the source to the generic space to be applied linguistically wherever the generic space is applied conceptually to a target space. But the projection of the generic space of cleansing to the particular target of wholesale killing of unwanted ethnic groups is not yet conventional, and the use of its vocabulary for this target is not yet the natural way to refer to the target. It follows that we are more likely in the case of “ethnic cleansing” to notice the projection and the generic space, and may even recognize the blended space.
Often, the blended space is forced into view. “Walking steam-engine” refers to a human being of great mechanical vitality, industry, direction, power, and seriousness. Many things underlying this phrase are entrenched: the projection of the space of mechanical devices to a generic space; the shift of vocabulary to that generic space; the wide application of that generic space to many targets; the consequent wide application of its vocabulary to those targets; and the conventional linguistic construction used for indicating that the target input space has a human story but the source input space has a nonhuman story, as in “walking encyclopedia” and “walking time-bomb.” Nonetheless, “walking steam-engine” evokes vivid specifics from the input spaces, impossibly blended. It is also an unconventional phrase. Accordingly, we notice the blended space.
In general, generic and blended spaces become more noticeable as the projection is less conventional, the expression is less conventional, or the projection and the expression combine noticeably what are still regarded as incompatible specifics. These circumstances are closely related: As the projection becomes conventional, so does the vocabulary it carries; and in the case of source and target input spaces, as the projection and the vocabulary become conventional, so the source category becomes extended, making the blend look less like a blend and more like an enlarged category. For example, the conceptual projections underlying “intellectual progress” extend the category of progress and journey to include mental events. Because this extension of the category is by now deeply entrenched, “intellectual” can be viewed as compatible with “progress,” and “intellectual progress” can be taken as referring to an extended part of the category rather than to an aggressive blend.
Even when the projection and its expression are most deeply entrenched and least likely to be noticed, generic and blended spaces are still available. Habit has simply made us take them for granted, rendering them invisible. The generic space projected from journey has its own existence independent of its embedding in the target story of the course of a life from birth to death. It can be projected to a different target altogether. We can project it onto a particular span of life, mental activity like problem solving, or the “wandering” characteristic of daydreams. We can project it onto marriage or artwork or the building of a bookcase or evolution or computer activity or any of a vast range of other targets. We can project it just as we project the generic spaces that arise out of “When the cat’s away, the mice will play” or “Look before you leap” or any other proverb that mentions only a specific source story. We can project it not just to purposeful behavior directed at a goal but indeed to any action performed by an intentional agent. “Watch your step” can be said of any action that an agent is about to perform, purposeful or not, intentional or not, consciously or not. The generic space underlying “Watch your step” consists only of an agent, an action that the agent is on the verge of performing, and a warning to pay attention.
A telling difference between a basic metaphor like LIFE IS A JOURNEY and our generic interpretation of “The girl who can’t dance says the band can’t play” is this: The generic space involved in LIFE IS A JOURNEY has a deeply entrenched default projection onto a particular specific target space, the course of a life from birth to death, while the generic space arising from “The girl who can’t dance says the band can’t play” is not conventionally tied to a specific target space. It is therefore easier to see the generic space in the second case, and to be aware of our application of that generic space to this or that specific target space.
When we take our data exclusively from deeply entrenched projections like LIFE IS A JOURNEY that have deeply entrenched vocabulary, the generic and blended spaces are less easily noticed, and the projection looks as if it carries positive meaning from one input space (the source) to another (the target). This has lead to the customary model of the projection of meaning as direct, one-way, and positive. This is a useful and parsimonious model, but it is adequate only in limiting cases.
Generic spaces differ linguistically from specific input spaces and blended spaces in one fundamental regard: They lack their own rich vocabulary. The vocabulary of a generic space is largely shifted to it from an input space. This vocabulary applies whenever we project the generic space onto a new space. For example, we have no generic word that means “an instrument somebody uses so constantly in his chosen work as to be regarded as defining the work.” One such instrument for one such worker is an axe. The relation of the worker to the axe—manipulating it, trying to get it to do what he wants done—has an abstract structure, and this structure can be projected to a generic space. That generic space can then be projected to a target, such as playing a jazz instrument. In jazz, someone’s instrument is called his axe. A saxophone is an axe, but so is a flute, a guitar, a drum set, a piano. The vocabulary of one space is shifted to the generic space, and projected from there to whatever target the generic applies to—in this case, jazz instruments. A new projection of the generic space onto a new kind of instrument (a synthesizer, for example) will not look unusual even if it is entirely novel: the vocabulary that has been shifted to the generic space is expected to apply to any target to which the generic space is projected. Such projection underlies category extension: “Axe” can now refer to the blended space that consists of all axes, including real axes.
We can at will focus on the generic space and wake it up fully. Even when, as in a basic metaphor, the generic space is fully superimposed upon a conventional target, we can nonetheless easily turn that generic space into a blended space by supplementing it with specifics from the source and the target. Underlying a phrase like “She’s making considerable progress as a young adult” is a conventional projection so deeply entrenched that it may be hard to see blending at work. The generic space, fully superimposed on the target, is nearly invisible. But this generic space can be turned into a blended space, as in a television commercial for an insurance company in which a person is actually on a road, encountering and going through doors labeled school, college admissions, undergraduate education, professional school, marriage. As she goes along, passing through each door on the road, she ages and her clothing changes to suit her present “stage of life.” She acquires fellow travelers in the form of playmates, friends, husband, and children. Her parents, standing at the side of the road, hand her fistfuls of money as she trots past. She in turn hands the money to the “bursar” who stands outside the door of professional school, and so on. This is a fantastic blended space, constructed by waking up the generic space and adding to it.
Waking up the generic space is a standard tool of literature. Contrast the very skeletal generic space underlying the phrase “drug trip” with Proust’s elaboration:
When one absorbs a new drug, entirely different in composition, it is always with a delicious expectancy of the unknown.… To what unknown forms of sleep, of dreams, is the newcomer going to lead one? It is inside one now, it is in control of our thoughts. In what way is one going to fall asleep? And, once asleep, by what strange paths, up to what peaks, into what unfathomed gulfs will this all-powerful master lead one? What new group of sensations will one meet with on this journey? Will it lead to illness? To blissful happiness? To death?
On n’absorbe le produit nouveau, d’une composition toute différente, qu’avec la délicieuse attente de l’inconnu Vers quel genres ignorés de sommeil, de rêves, le nouveau venu va-t-il nous conduire? II est maintenant dans nous, il a la direction de notre pensée. De quelle façn allons-nous nous endormir? Et une fois que nous le serons, par quel chemins étrangers, sur quelle cimes, dans quels gouffres inexplorés le maitre tout-puissant nous conduira-t-il? Quel groupement nouveau de sensations allons-nous connaître dans ce voyage? Nous ménera-t-il au malaise? À la béatitude? À la mort?
Similarly, contrast the very skeletal generic space underlying an expression like “Death finally won” with Proust’s elaborations:
For we talk of “Death” for convenience, but there are almost as many different deaths as there are people. We do not possess a sense that would enable us to see, moving at full speed in every direction, these deaths, the active deaths aimed by destiny at this person or that. Often there are deaths that will not be entirely relieved of their duties until two or even three years later. They come in haste to plant a tumour in the side of a Swann, then depart to attend to other tasks, returning only when, the surgeons having performed their operation, it is necessary to plant the tumour there afresh.
Car nous disons la mort pour simplifier, mais il y en a presque autant que de personnes. Nous ne possédons pas de sens qui nous permette de voir, courant à toute vitesse, dans toutes les directions, les morts, les morts actives dirigées par le destin ver tel ou tel. Souvent ce sont des morts qui ne seront entiérement libérées de leur tâche que deux, trois ans aprés. Elles courent vite poser un cancer au flanc d’un Swann, puis repartent pour d’autres besognes, ne revenant que quand l’opération des chirurgiens ayant eu lieu il faut poser le cancer à nouveau.
In such cases of rich blending, the generic space is awakened and elaborated in a blended space where new work is done that requires revision of the input stories.
A generic space stands at one end of a gradient; a richly blended space stands at the other end; the blend connects to the generic space and includes its structure. As information is blended from the input spaces into the blended space, it moves along the gradient from generic to blend. The degree of blending varies along this gradient. A “pedagogical gun” or a “walking steam-engine” may suggest a thin blend, while a lengthy allegorical poem may suggest a very rich blend.
The degree of blending is often up to the reader. Bernadine Healy resigned as head of the National Institutes of Health (NIH) shortly after the inauguration of President Bill Clinton, implicidy acknowledging that she had been asked to leave because she had been associated with the federal ban on the use of fetal tissue in scientific research paid for with federal funds. She expressed regret that the agency had been drawn into the political debate by saying, “NIH has become a bit of the Beirut of abortion and fetal tissue.”
A reader who uses the popular notion of Beirut as suffering terribly and innocently because conflicting factions war over it might interpret this blend so thinly as to make it seem like no blend at all: Healy is simply saying that the NIH has suffered innocently by being caught in the “political crossfire,” where again the noun comes from the source and the adjective from the target but in a phrase so conventional as to seem as normal as “intellectual progress.”
But it is equally easy to imagine Healy in a bullet-scarred NIH building (the one in Bethesda, Maryland, just outside the District of Columbia) taking political rounds of ammunition, lobbed at her alternately by Republican and Democratic presidents and their political factions, from the White House and Congress, located on the Mall a few miles away. A political cartoon supporting her point of view might picture her in a helmet, ducking the incoming fire. There is a gradient of specificity between the generic space and the blended space, and we have latitude in moving along that gradient as we interpret expressions.
Conceptual blending is a fundamental instrument of the everyday mind, used in our basic construal of all our realities, from the social to the scientific. Let us consider some examples of social and scientific blending originally treated by Fauconnier and me.
Analogy places pressure upon conventional category structures. A successful analogy can, through entrenchment, earn a place among our category structures. The assault of an analogy on conventional categories is often expressed in the early stages by a blend-construction that draws its noun from the source and its modifier from the target. “Same-sex marriage,” for example, asks us to project the scenario of marriage onto an alternative domestic scenario. People of violendy opposed ideological belief will freely agree that the generic space of this projection carries information applicable to both scenarios. It might include people living in a household, dividing labor, protecting each other in various ways, and planning together.
What is at issue is not the existence of this information but rather its status. Those whose conception of conventional marriage has as a requisite component “heterosexual union” and has as its prototype “for the sake of children” will regard this abstract generic information as merely incidental or derivative in the story of traditional marriage. In their view, “same-sex marriage” will remain an analogical projection whose blended space is as fantastic and conflicted as any we have seen in Dante or Shakespeare, but which nonetheless draws legitimate abstract connections—however inessential—between one kind of story and an entirely different kind of story.
But others may regard this abstract information in the generic space as the central information in the story of traditional marriage. They may regard “heterosexual union for the sake of children” as merely incidental information in the traditional story. For these people, “same-sex marriage” is not an aggressive analogical construction; it simply refers to a subcategory of marriage in the way that “light wave” refers to a kind of wave.
We must be careful not to mistake the research question of how parable works in the mind for a different issue of professed ideological belief. Agreeing to treat two scenarios as belonging to the same category for purposes of protection under the law or taxation or health coverage or whatever is different from actually having a conceptual structure in which these two scenarios belong to the same conceptual category. Liberal goodwill toward diverse scenarios, on a philosophy of live and let live, is irrelevant to the phenomenon of recognizing something to be an obvious instance of a category, as when we recognize light to be obviously a wave or a heron to be obviously a bird. When we have a category that is entrenched in our conceptual structure, we do not merely agree to treat two of its central members in some of the ways we treat the other, instead, they share the same default generic structure.
Suppose the information in the generic space of “same-sex marriage”—people living in a household, dividing labor, protecting each other in various ways, and planning together—came over time to be the central information in our standard concept of marriage. This would mean that the generic space of “same-sex marriage” had displaced the generic space of “traditional marriage” as the essential structure of the category marriage. The blended space would become not a fantastic combination but rather a new and wider category. It would ultimately subsume the original input spaces.
In a situation of ideological tension over category connections, there is always the opportunity to reject the projection and invert the relative status of the two input spaces. This has happened in debates over racial, gender, social, and economic categories, and occurs in two steps. First, an attempt is made to claim that the generic space that is projected from a powerful group (white, male, aristocratic, rich, colonialist, capitalist, master) applies to less powerful groups if we project the “right” and “just” information, thereby extending the category: There becomes no “legitimate” distinction between aristocrat and commoner, for example. Such projections are often resisted by those who see themselves as belonging to the source category, but embraced by people who see themselves as belonging to the target.
However, a second possible response in the social debate occurs when those in the target reject the projection, on the claim that the two groups do not belong to a category at all. It might be claimed, for example, that violence is central to men but not to women and that any generic space that lumps them together does a disservice to women and is to be rejected; that the rich are inherently dishonest while the poor are inherently honest and that any generic space that lumps them together does a disservice to the poor and is to be rejected; that white culture is essentially a culture of ice and therefore cold while black culture is essentially a culture of the sun and therefore warm and that any generic space that is proposed as a category that applies to both of them is to be rejected; that conventional marriage involves asymmetry between the man and the woman while same-sex marriage has no such asymmetry and so any generic space proposed to lump them together is to be rejected. What is at issue here is of course not in the slightest degree any particular ideological view but rather the fact that all ideological views use parable to judge and reason. Parable is an instrument of thought and belief and consequently of argument.
The cultural tussle over the analogical pressure of “same-sex marriage” upon conventional category structures provides daily journalistic copy and stirs passions. It is an example of the role played by blended spaces in our understanding of cultural and social reality, and of our place in that reality. Blended spaces play the identical role in the world of basic science. Consider the case of “artificial life.” If a mental space that includes biological life has as central information “embodied, developed through biological evolution, carbon-based,” and so on, then “artificial life”—which comes from a computer lab and is not based on carbon—will always be an analogical concept, and “artificial Ufe” will not belong to the category “life.” It will be a provisional category extension, like “He’s a real fish.” But computer viruses, for example, share abstract structure with biological organisms. As the generic space that can be projected from biological life and imposed on computer events grows more useful, some people may be tempted to change their conception of the status of this information as carried in the source. The generic space involved in the concept “artificial life” could in principle come to constitute the central structure of the source. In that case, “artificial life” would become a subcategory of life. At present, “artificial life” is an analogical projection of evident utility that seems unlikely to displace conventional category connections. But that situation could in principle change.
Blended spaces play a routine role in the development of even the most fundamental scientific concepts. Mass and energy, once conceived as belonging to two different categories, have been reconceived. The blended space of mass that is simultaneously energy has become the new category: Mass is energy and energy is mass. Similarly, the projection of spatial structure onto the conception of time has always been profound; in our own century, the scientific role of this projection has grown considerably larger, resulting in a blend that seemed at first an impossible clash—space-time. But this blend too has come to subsume the original inputs.
There is a certain way of thinking about thought that has kept blended spaces from being detected. It begins with the ostensibly reasonable assumption that inference and truth go together. On this assumption, central meanings and crucial inferences that guide our action come from what we believe to be true, not from fantasy constructions like Bertran de Born in Dante’s Inferno. To get over this blindness to blended spaces requires separating belief from the construction of meaning. It is often the case that central inferences for a “real-world” space are in fact constructed in a blended space that we do not believe to be true or real. The truth of an input space can comefrom a blended space that we do not believe to be true or real. This sounds paradoxical, but there is no reason it should, once we draw a distinction between the construction of meaning and the adoption of belief.
For example, personification—or even the minimal projection of intentionality according to EVENTS ARE ACTIONS—results in a blended space whose “truth” we hold quite distinct from the “truth” of the target space, yet inferences are clearly projected from the blended space to the all-important target space. In Hemingway’s The Old Man and the Sea, the fisherman’s hand cramps. He speaks to his hand as if it is his enemy in the struggle to bring in the fish. He taunts it. The fisherman is not demented. He does not for a moment actually believe that his hand is an intentional adversarial agent. Relative to the blended space, it is true that his hand is an intentional adversarial agent. Relative to the target space of his fishing and the work of his body in that scenario, it is not true that his hand is an intentional adversarial agent, and he has not the slightest confusion regarding the difference. He keeps the two spaces quite distinct. Nonetheless, there are certain inferences from the blended space that he does project to the target space and believe of that target space, most salient among them that he will do best to adopt the emotions and strategy of the blended space. This is an inference that bears on the target space: He, the undemented fisherman, should in the target space of what he actually believes apply himself to dealing with his hand as if it were his intentional opponent while understanding fully the weight of as if. He should work to an extent in the blended space, without believing it. Useful construction of meaning is not the same as adoption of belief.
This is the stance we all take when we are dealing with a recalcitrant problem or task and construct the extraordinarily useful blended space in which this problem or task is our opponent—a tax fiddle to be worked, a tire to be changed, a tent to be staked down. If in analyzing the construction of inference we restrict our focus to what we really believe to be true, we will be blind to the indispensable blended spaces in these cases. If we ask someone cursing at a tire she is trying to change whether the tire really has an intention to frustrate her, she will think we are making a joke. In the all-important target space, she understands the tire as not intentional. But her useful attitude depends upon a blended space she does not believe.
On the other hand, a blend need not be considered false or fantastic. It can come to be regarded as compatible with the source. “Working mother,” “provisional government,” “independent scholar,” and so on might be explainable as cases where one general narrative (mothering, government, scholarship) is combined with another (working, provisional activity, independent activity) into a blended space. The hypothesis that the linguistic construction represents conceptual blending is radical. It is traditional to assume that the modifier-noun construction represents propositional expression (“red shoe”) but can be adapted or stretched for exotic purposes like blending (“Irish twins”—said of two siblings born less than twelve months apart). The radical view reverses the traditional direction: The modifier-noun construction represents blending, but in special minimal circumstances, the blending can become invisible, misleading us into thinking we are dealing only with propositional structure. Consider “fire station.” This is a very common noun phrase that evokes no surprise. But if we think about it, a fire station does not have fire, provide fire, or receive fire; fire is not part of it or the category that includes it. In fact, fire is in no way a feature of a fire station. But we have one general story of fire and another general story of people and equipment being stationed at a station for a purpose, and we can blend them into a story in which fire is not a feature of the station or a counterpart of the station. “Fire station” asks us to do just this.
Charts, diagrams, coins, and maps frequendy provide visual representations of blended conceptual spaces. Consider the projection from magnitude of size to magnitude of importance, as in “Let’s attend to the larger issues.” A blended space is available in which the more important things appear spatially larger. A tourist map of a city may include perspectival drawings of major attractions that are inset here and there in the map—the more important the real attraction, the larger its inset picture, regardless of the relative physical size of the actual attraction. A map to literary sites of Paris might feature a drawing of Hemingway’s tiny apartment that dwarfs the drawing of the huge Arc de Triomphe.
Maps are visual representations that bring together information from many different mental spaces associated with what the map represents. These spaces include street topography, names of places, latitude and longitude, appearance, water versus land, distances, directions, number of hours it takes to travel from city to city, quality of food, availability of services, hours during which events take place or buildings are open, dates of construction, weather, temperature variation, altitude, and so on. The fact that the map gives simultaneous access to these very different spaces does not automatically indicate that there is mental blending at work. An outline of the landmass of the United States filled in with colored bands to indicate temperature ranges combines spatial position and color so as to give us access simultaneously to the mental space of geographical location and the mental space of weather, and it gives us a mental space connector from the mental space of geographical location to the mental space of weather; but it does not signify a mental blend in which geographical locations are simultaneously colored. The combination in the map signifies mental space connection, not mental space blending. However, such representations of mental space connections will recruit conventional blends whenever possible. For example, there is a conventional mental blend of temperature and color (hotter is redder), and it would be infelicitous to contravene that blend in the visual representation, using for example red to indicate cold and blue to indicate hot. Maps are an interesting case: They are representations that give us access to many mental spaces. Combinations of symbols on the map indicate mental space connections and correspondences. But these maps also slide, where possible, into visual representations of mental blends.
Visual representations of mental blends are often immediately intelligible. We seem to recognize their meaning passively, as if they pose no puzzle and require no work. However, once analysis detects the inferential role of the blend, it can be surprising to see how much work we actually did to interpret the drawing. The New York Times Magazine of August 22, 1993, carries an illustration by Brian Cronin of an article by Thomas L. Friedman, “Cold War without End.” In this illustration, a bicyclist is riding a bicycle; the front half of the bicycle frame is blue with a small American flag attached; the front wheel is the globe of the world, with lines of latitude and longitude; the back half of the bicycle frame is a red hammer and sickle—the sign of Soviet political control. The back fender is the sickle blade, and the pedal is the sickle handle; the strut to the center of the rear wheel is the hammer. But there is no rear wheel! The rider is looking at the place where it should be. He is understandably dismayed.
The illustration accompanies an article that suggests that the makers of American foreign and domestic policy are troubled over how to proceed now that Soviet control has disappeared. The foundation of American policy had been conflict between the United States and the Soviet Union. That foundation is gone, and nobody seems to know what to do. The article depends on the metaphor that the making of policy is bodily movement; that the cold war was an instrument for bodily movement. This metaphor is an instance of the universal projection of stories of body movement onto stories that are not bodily or even spatial. It is clear to anyone looking at the bicycle illustration that the vehicle is flawed and that normal progress cannot be made. This inference can arise in the source space of bicycling—a bicycle without a wheel won’t go. It is also clear that something that previously existed has disappeared in a dismaying fashion. This inference can arise in the target space—Soviet control over republics and satellites has simply vanished. But these two conceptual structures from the source and the target are blended into a fantastic and impossible space.
The central inference of the illustration is that Soviet control has disappeared in a way that automatically and inevitably brings a grave and manifest crisis to America, and that something must be done immediately, before the whole show comes crashing down. This inference is impossible for the source space of bicycling. Magic aside, a bicycle wheel cannot simply disappear, leaving the rider suspended in midair for a vertiginous moment that lasts long enough for him to notice the missing wheel and feel dismay. The inference is not necessary in the target—indeed, it is so counterintuitive to imagine that the disappearance of Soviet political control should be bad for American policy making that the writer must spend the entire article trotting out evidence. Yet in the blended space, the wheel that is Soviet dominion can disappear, and the disappearance of the wheel that is Soviet dominion must have inevitable, unmistakable, and incontrovertibly bad effects upon the bicycle that is the vehicle of American foreign policy. The central inference is constructed in the blended space and projected from there to the target.
Every newspaper and news magazine carries editorial columns. A cursory survey of them will reveal that a favorite routine for writing a column is to churn out a blended space, often labored. Once we see how this works, it is very easy to do, as we can show by fabricating an example. At the end of President Bill Clinton’s first one hundred days in office, The New York Times reported that Clinton seemed unable to accomplish anything in Washington, D.C., despite his promise to hit the ground running. Not a single undersecretary had been appointed. Senator Bob Dole of Kansas, the ranking Republican on the hill, had just handed President Clinton a major defeat on a jobs bill. Clinton had lacked four votes in the Senate and had seemed ignorant of how to get them. New York Times columnist Russell Baker, not a fan of the often abrasive Dole, compared Clinton to President Jimmy Carter, saying essentially that a president who doesn’t know how to use the powers of his office to buy four lousy votes in the Senate is simply incompetent—this would never have happened to President Lyndon Johnson.
Russell Baker frames Clinton and Dole as opponents, and that is all we need, since it gives us a generic space for opposition and a target space of Clinton and Dole’s political opposition. A vivid source space might be a gunfight. In the blended space, Bob Dole becomes the Kansas gunslinger with the big iron on his hip, who, like his counterpart in the movie Shane, is cool, confident, utterly professional, wastes no energy, picks his fights, and knows his ground. Something like this: “Dole is the ranking gunslinger in Dodge-City-on-the-Potomac, rough at the edges but smooth on the legislative draw. When the sun of deficit negotiations got in Clinton’s eyes, Dole made his move and Clinton never cleared leather. Hillary Clinton is sure to come running out with her health-care initiative to try to plug the wound, but Sheriff Clinton is bleeding his heart out on Main Street U.S.A., and Doc Greenspan at the Fed, with a barely recovering economy to protect from inflation infection, isn’t about to remove the bullet so Clinton can pull himself up and go spend another ton of money.…”
It is not only on the editorial page that such blends are allowed. A lead sentence on page 1 of The New York Times for May 13,1993, reads, “An unexpected surge in wholesale prices last month, the latest of a string of higher price reports, left many economists and investors wondering whether the inflation genie was starting to slip out of the bottle.” Immediately after reading this newspaper, I called United Airlines and received the recorded while-you-wait-for-an-agent message, “Now you can tango on down to Rio, with less fancy footwork”—presumably an ad for a new nonstop flight. “Tango” and “fancy footwork” come from the source space but “on down to Rio” comes from the target space, since intercontinental travel is not possible by means of tango. I am told that the phrase also blends in elements of a Fred Astaire movie about flying on down to Rio.
In 2 Samuel 12, the prophet Nathan comes to King David. David is a judge, and Nathan approaches David in that role. Nathan presents David with the case of a certain rich man and a certain poor man.
The rich man had very many flocks and herds; but the poor man had nothing but one little ewe lamb, which he had bought. And he brought it up, and it grew up with him and with his children; it used to eat of his morsel, and drink from his cup, and lie in his bosom, and it was like a daughter to him. Now there came a traveler to the rich man, and he was unwilling to take one of his own flock or herd to prepare for the wayfarer who had come to him, but he took the poor man’s lamb, and prepared it for the man who had come to him.
Then David’s anger was greatly kindled against the man; and he said to Nathan, “As the Lord lives, the man who has done this deserves to die; and he shall restore the lamb fourfold, because he did this thing, and because he had no pity.”
Nathan said to David, “You are the man.”
The projection must be immediately obvious to David. David had been anointed king over Israel and enjoyed many wives. Had this been too little, the Lord would have added to him as much again. But Uriah the Hittite, one of David’s subjects, had only his one wife, Bathsheba. David took Bathsheba for his temporary pleasure, and she conceived a son by David, and David arranged for Uriah to be placed in the front ranks of the battle and abandoned, so as to be killed. The judgment proclaimed by David upon the rich man comes down upon his own head.
In telling his story, Nathan has pretended that it is the space of the rich man and poor man that is the target, and the space of family domesticity and affection that is the source (the ewe lamb “was like a daughter to him”). In doing so, Nathan has led David to construct a strong blended space that contains specifics of both the story of the rich man and the poor man and the story of relations between members of a family (especially “eat of his morsel, and drink of his cup, and lie in his bosom”). David thinks he knows where all this information is directed, and why: It is to clarify the iniquity of the rich man and ensure his condemnation. But then Nathan announces that the target of the projection of this blended space is not the story of the poor man and the rich man, but the story of David, Uriah, and Bathsheba.
This is veiled parable: In order to prevent the listener from resisting the projection, the storyteller veils the intended target while building up a blend with the right structure for his real purposes. In the case described by Nathan, that blend includes family affection, relations of power and its abuse, and categories of just and unjust behavior. Once this slighdy blended space is fully constructed, Nathan lifts the veil from the real target and conjures David to project inferences from this blended space to it. The fit is extraordinarily compelling: The final target strongly resembles the source—both source (“it was like a daughter to him”) and final target (Uriah, Bathsheba, and David) concern the destruction of a family. The establishment of counterparts has been so carefully developed by Nathan that David has no escape.
An emblem is a parable that starts from one story and projects from it a generic story that covers other stories belonging to the same conceptual domain. A story about a boy’s adventures at summer camp can be interpreted as an emblem of childhood adventure. I have interpreted the story of Shahrazad and her father the vizier as an emblem: we project from it a common and abstract story of confrontation between child and parent. That generic story applies to many other stories belonging to the same domain as the original story.
When the generic space applies to a story in a different conceptual domain, we have metaphor or analogy instead of emblem. This reveals something interesting: Whether a given parable is emblematic or metaphoric depends upon what constitutes a division between conceptual domains, which is to say, upon what conceptual connections are already in place when the parable becomes active.
For example, The Thousand and One Nights begins with a story that can be read as either emblem or metaphor. King Shahzaman is called from his kingdom by his elder brother King Shahriyar to pay a visit. Shahzaman departs, leaving his kingdom in the hands of his vizier. But out of deep love for his wife, he finds a pretext for returning, so he may see her once more. He discovers her making love to a slave and kills them both instantly before departing on his journey. He is desolate when he arrives, declines to reveal the cause to his brother Shahriyar, and remains behind in the palace when King Shahriyar goes out to hunt. Accidentally, Shahzaman witnesses a marvelous sight: His brother Shahriyar’s queen enters the courtyard with twenty slave-girls, ten of whom turn out to be slave boys who make love to the other ten slave girls, while the queen calls to herself a special slave to do the act. Shahzaman realizes he is not the only one to suffer. His mood improves. That night, Shahriyar, recognizing the change in his brother, manages to extract from him the reason for his initial despondency but not for his recovery. But at last, Shahzaman tells his brother everything. Alarmed, filled with doubt, Shahriyar spies on his wife, discovers the truth, and “half demented at the sight,” says to his brother, “Let us renounce our royal state and roam the world until we find out if any other king has ever met with such disgrace.”
They roam. One day, while the brothers rest under a tree, a black pillar rises from the surging sea until it reaches the sky. The terrified brothers climb the tree and watch as a giant jinnee carries a chest on his head. He places it on the shore below the tree, opens the chest to remove a box, and then opens that box to take out a “beautiful young girl, radiant as the sun.”
The jinnee falls asleep in the girl’s lap. She sees Shahriyar and Shahzaman and signals to them to come down and make love to her, on threat of her waking the jinnee to rip them apart should they decline. Afterward, she takes their seal rings, for a total of one hundred rings so acquired.
“This jinnee,” she added, “carried me away on my bridal night and imprisoned me in a box which he placed inside a chest. He fastened the chest with seven locks and deposited it at the bottom of the roaring sea. But he little knew how cunning we women are.”
The two Kings marvelled at her story, and said to each other: “If such a thing could happen to a mighty jinnee, then our own misfortune is light indeed.” And they returned at once to the city.
As soon as they entered the palace, King Shahriyar put his wife to death.
Misery loves company. The two kings read this story emblematically: Women are always cunning and deceitful. From the specific story with a specific woman and a specific male who tries to keep her, they project an abstract generic story: The female dupes the male through her cunning. This generic story is then available for projection to any other particular space, such as the specific stories of both Shahzaman and Shahriyar. The kings have a psychological interest in reading the story this way, as an emblem that provides a generic story of what women do to men.
There is of course a different projection available to us that it is psychologically impossible just now for the kings. The jinnee, serving himself, exerts extraordinary control over the young woman, “taking” her physically from one location to another on her wedding night, limiting her freedoms in specifically spatial ways. His behavior causes her to hate him, which leads her to seek revenge: She sets out to do exactly what he has so single-mindedly tried to prevent her from doing. Given this causal sequence, we can regard the jinnee as the cause of his own predicament and blame him for it. This story can be projected metaphorically onto the story of marriage generally, or at least, certainly, of marriage to kings. “Stealing” the woman like an object on her wedding night, regardless of her desires, can be projected metaphorically onto what kings do with their brides. The jinnee’s locking her up and keeping her under the sea can be projected onto the kings’ placing limitations (spatial and nonspatial) on their wives—Shahriyar’s wife had to stay in the court while he went out to hunt. This projection is not available to Shahzaman and Shahriyar at this point: it would imply that they and not their wives are to blame for their being up a tree.
Grammatical constructions often represent basic abstract stories. For example, consider the basic abstract story of result: An intentional agent performs an act on an object, with a particular result for the object. This basic abstract story is represented by the grammatical resultative construction, used in “Kathy pressed the wallpaper flat” and “Mike painted the kitchen yellow.” Consider also the “caused motion” construction studied by Goldberg. It has the syntax
NounPhrase—Verb—NounPhrase—Preposition—NounPhrase,
as in “He hit the ball into the bleachers” or “They shoved the poor guy out of the room.” The grammatical construction represents a basic abstract spatial action-story in which an agent performs a physical action that causes a physical object to move in a spatial direction. Fauconnier and I have analyzed the way in which this grammatical construction serves as a prompt to blend two things: the events referred to by the vocabulary filling the construction and the caused-motion story represented by the construction itself. In the resulting blend, the unintegrated events to which the vocabulary refers acquire the integrated event structure of the caused-motion story. For example, if we say, “He floated the toy boat across the pond,” then even though float encodes only the manner of the boat’s motion, and not the causal action or the causality between that action and the motion of the boat, we still know that we are being asked to conceive of a scene in which the floating of the boat is a story of caused motion. As Goldberg observes, “He sneezed the napkin off the table” has a verb that is a “parade example” of intransitivity, but its use in this construction prompts us to think of a scene in which sneezing physically causes the motion of an object in a direction. It does not mean, for example, that he sneezed and someone picked a napkin up off the table to hand him, because that is not a caused-motion story. We blend the physical caused-motion story with the story of sneezing.
We project from the basic abstract action story of caused physical motion an even more abstract generic story of causation that is not necessarily physical and that can be projected to target stories that may not be physical. For example, we can say, “They laughed the poor guy out of the room,” even though laughing is not a physical force. We can say, “She drove him out of his mind” even if she exerted no physical force on him and even though the result is not physical motion in a spatial direction. The basic abstract story of physical caused motion involves physical spatiality; the generic story of caused motion does not, and can apply to targets that are not physical or spatial.
Grammar contains many devices for indicating what mental spaces are to be built and how they are to be connected. These devices include tenses, moods, some verbs, and conjunctions. Grammar is a set of instruments used to guide the building and blending of spaces and the projecting of generic spaces. In other words, grammar is the servant of parable. In the final chapter of this book, we will consider why it is such an apt servant.
As an example of how grammar gives us guidance in space building, space blending, and projection of generic spaces, consider the xyz construction, whose syntax is deceptively simple:
NounPhrase(x) be NounPhrase(y) of NounPhrase(z)
as in Vanity is the quicksand of reason. This simple syntax has a complex semantic and pragmatic interpretation: Construct a set of spaces and projections with the result that x in a target is the counterpart of y in a source, and z in that target is the counterpart of an unmentioned fourth element ω in that source. In “Vanity is the quicksand of reason,” ω is the traveler who travels toward a goal. As quicksand stops the traveler, vanity stops reason. The grammatical information is minimal and highly abstract—find a mapping and a missing element. The rest is left to the cognitive competence of the user, who must construct an implicit generic space and an implicit blend, in which a single element is simultaneously reason and a traveler.
An abbreviated form of the xyz construction can be seen in phrases like “Language is fossil poetry.” Emerson meant by this expression that language, which seems so literal, consists of historical traces of inventive and original “poetry.” Here, x is language, z is poetry, and y is given by the modifier fossil. A missing W (living organism) must be found to complete the correspondences. We construct a blended notion of something that is simultaneously language, poetry, and fossil: in the blended space, poetry is a living organism that can leave a fossil trace (language).
Noun phrases of this yz form (“fossil poetry”) arise often. We call a suburban teenager a “mall rat.” We refer to a “tax bite,” a “computer jock,” a “price war,” a “surfing safari,” or “channel surfing (or internet surfing).” The construction can be used to signal new blends, as when the Clinton administration’s comprehensive effort to make executive appointments along racial and gender guidelines was called by a Republican congressional staffer a “diversity jihad.” A related construction uses the standard adjective-noun syntax to indicate counterparts: “In this way he acquired a vast hoard of all sorts of learning, and had it pigeonholed in his head where he could put his intellectual hand on it whenever it was wanted,” or, “In making these investments, you are digging your financial grave.” Someone who has left the Touraine and its cuisine to return to the American Midwest might be said to suffer upon reentry “the gastronomic bends.” A Parisian who dates one woman only on the Right Bank and another only on the Left refers to this as “French fidelity.” Universities that try to raise money for research by making arrangements for technology transfer are said to be holding “technological bake sales.”
There is an aspect of parable that so far we have acknowledged only implicitly. It concerns gradients of spaces and projections. The basic metaphor DEATH IS DEPARTURE, for example, operates at one level of specificity, but above it lies EVENTS ARE ACTIONS, which projects information even more abstract, and above EVENTS ARE ACTIONS lies the most general level of projection, the general capability of projecting one story parabolically onto another.
Below DEATH is DEPARTURE on the gradient of specificity lies the projection of the story of an agent carrying somebody off forever onto the story of death. Below that lies the projection of the story of Thanatos intending to snatch the body at the grave and carry it offto Hades, but being wrestled out of it by Heracles onto the story of the near death of Alcestis, and the projection of someone trying to pull Alcestis by the hand toward Charon the ferryman who is waiting to carry her across the river Styx onto the near death of Alcestis. A projection at any level in such a graded hierarchy can provide structure to any other projection in the same hierarchy.
Parable involves dynamic construction of input spaces, generic spaces, and blended spaces, multiply linked, with projections operating over them. Inferences and meaning are not bounded in a single conceptual locus. Meaning is a complex operation of projecting, blending, and integrating over multiple spaces. Meaning never settles down into a single residence. Meaning is parabolic and literary.
This view of meaning may seem counterintuitive. If we think of concepts as litde packets of meaning whose boundaries circumscribe our knowledge of them, then any sort of meaning we recognize as distributed over many spaces may seem secondary, marginal, special, or parasitic. On that view, a blended space will seem to be an exotic mental event, an exceptional quirk. But the variety of the examples we have seen shows that the dynamism, distribution, projection, and integration we see in blending are actually central and pervasive elements of everyday thought.
Consider any basic everyday concept, like house. It seems static and permanent, stable and unitary, cohesive and self-contained. This is an illusion, derived from an influential but mistaken folk notion of concept. We have no concept house, but we do have a word “house,” and being able to use that word—like the words “fish,” “marriage,” “life”—requires us to construct, activate, link, and project the appropriate configurations of spaces, frames, and cognitive models. We may not perceive this multiple activation for a word like “house” because we activate again and again the same configurations of spaces and links for “house,” so that one of those configurations can come to seem to be our “basic concept house.” But in fact a great range of spaces is activated for “house”: shelter from the elements, bounded interiors, security from intruders, financial investment, artifacts, aesthetic design, instrumentality for inhabitants, social residence, social place of meeting, partitioning of activities into different physical spaces, rental property, and on and on indefinitely. Any single use of the word “house” for any particular purpose will involve construction of meaning as an operation of selective integration over these various distributed stories.
Let us look at a case where it is clear that different spaces are activated for the same word in different local situations. Consider the simple xyz equation, “Italian is the daughter of Latin.” The source input space contains overtly only a very little structure: It has a parent and a daughter. The target input space also contains overtly only a very little structure: It has Italian and Latin. Ultimately, parent in Úas source will correspond to Latin in this target, and daughter in this source will correspond to Italian in this target. The result is a double personification: In the blend, Latin and Italian are human beings. But this hardly accounts for the meaning we construct on hearing this phrase. One normal interpretation might be paraphrased: “Latin existed first, and Italian came into existence second by deriving causally from Latin.” It may seem as if the material needed for constructing this meaning is intrinsically in the source input space of parent-child and the target input space of Latin-Italian. But it isn’t. It must be recruited and brought into the source input and the target input. It can be recruited by activating other spaces and projecting them to the skeletal source and target inputs. Onto the skeletal target input, we project a space concerning languages. In it, some languages derive from others, and in particular, Italian derives diachronically from Latin. Onto the skeletal source input of parent and child, we project a particular story of progeneration. In it, a parent bears a causal and sequential relation to the child; the parent produces the child; the parent precedes the child.
If the source input and the target input are elaborated in these ways, by this recruited structure, they share a causal and sequential structure. This causal and sequential structure will then be available to constitute the generic space that underlies the interpretation “Latin precedes and results in Italian.”
But now suppose that we are talking about the study of foreign languages in a particular high school, and we observe that nearly all the students of Latin come to the subject because they encountered it in courses in Italian. We might say, “Italian is the mother of Latin.” Does this assertion contradict the earlier assertion that “Italian is the daughter of Latin”? Clearly not. In the case of “Italian is the mother of Latin,” what has been additionally recruited to the target input is the sequential order in which languages are learned, not the order in which they historically derived. In the space of learning languages, the sequential order of Latin and Italian is the reverse of their sequential order in the space of the history of languages. Recruiting structure from the space of learning languages to the space of Italian and Latin lets us say, “Italian is the mother of Latin,” while recruiting structure from the space of the history of languages to the space of Italian and Latin lets us say, “Italian is the daughter of Latin.” In general, there is no fixed structure of the target input space that the source input space must match, because the target input space has different structure under different recruitments to it.
Now let us consider the source input space. Suppose we are discussing the relative aesthetic qualities of Italian and Latin, and we comment upon the precision of Latin in Vergil, Propertius, or Horace as compared to the florid and ostentatious qualities of Italian in Bocaccio or Tasso. We say, “Well, Italian is the daughter of Latin, and her ostentatious beauty is really a rebellion against her mother’s austerity.” Here, we recruit to the source input space of mother and daughter not just progeneration but also social relations between mothers and daughters, and in particular adolescent rebellion over appearance and behavior. Under this recruitment of structure to the source input space, Italian is still the daughter of Latin, but in an entirely different additional sense.
In all these examples, the underlying conceptual domains—kinship and languages—are the same, but the spaces selected for recruiting additional structure to the source input space differ example to example, and the spaces selected for recruiting additional structure to the target input space differ example to example. The resulting configurations—source input space, generic space, target input space, and blended space—are different in each case.
Some recruitments to source and to target will be more common than others and thus may come to mind more quickly and give rise to highly salient default interpretations. In isolation, “Italian is the mother of Latin” will raise eyebrows. Common or default recruitments are a phenomenon of thought in general: we are always ready to use default conceptual connections as we think. It is important to recognize, however, that common, default recruitments do not give us fixed basic concepts: we can always unplug the default connections; they are, in technical jargon, “defeasible.” They look stable and fixed sometimes, but only because they are entrenched. Our most entrenched concepts and connections are formed by the same mechanisms of parable we have seen in the exotic and unusual cases.
By now, we have detected many features of parable: input spaces (which sometimes are related as source and target); abstract structure that is shared by inputs; generic spaces that contain that shared structure; counterpart connections that exist between inputs because of their shared structure; projection from inputs to a blend; development of emergent structure in the blend; projection of structure, inferences, and affect back from the blend to inputs; and variable recruit-, ment of structure from other spaces to the inputs themselves.
This many-space model of parable makes it possible to give a concise statement of the invariance principle: Conceptual projection, which has as one of its fundamental activities the projection of image-schematic structure from a source input to a target input, shall not result in an image-schematic clash in the target. Invariance is a global constraint to be satisfied in building and projecting target, generic, and source spaces.
One corollary of the invariance principle is this: Once the projection is completed, the most abstract generic space, the one that contains just the image-schematic structure taken to apply to both source and target, shall not contain an image-schematic clash.
There are many ways to avoid such a clash. We have choice in what we recruit to the source, what we recruit to the target, and what we project to the generic space. We can vary all of these choices in order to meet the constraint. For example, “Italian is the mother of Latin” can be interpreted as meeting the constraint so long as we recruit to the target not the space of historical derivation of Romance languages but rather the space of order oflearning languages in a particular school.
There is one way to meet the constraint that may not be obvious. Suppose a (crackpot) professor of Romance languages says, “Italian is the mother of Latin,” and we respond, “You can’t mean what I think you mean, you can’t mean that Latin as a language derives from Italian as a language,” and the professor responds, “Oh, yes, I do. Originally there was a small tribe of Etruscans who developed a new language, and it was Italian. Latin really derived from that language, although nobody recognizes this since the original speakers of Italian couldn’t write—they just carved wooden sarcophagus lids all the time, and this took up such a ton of energy every day that they were always too tired to get around to learning the alphabet—so they left no documents. Later on, of course, Latin turned back into Italian, but in fact Italian never stopped being spoken up in the Apennines; it was just never written down. So you see, everyone thinks Latin is the mother of Italian, but as a matter of historical fact, Italian is the mother of Latin.” In this case, we are being directed to meet the constraint of the invariance principle not by changing projections to source, target, or generic space, but rather by erasing and re-forming the image-schematic structure in the conceptual domain historical derivation of Romance languages so that when its structure is recruited to the target, the result will not clash with the image-schematic structure we hope to project from source to generic to target. We are unlikely to do this. Instead, we will build a separate space of what this professor believes about Latin and Italian, and we will note that it does not agree with what we believe.
The mental ability to combine and blend concepts has always seemed exceptionally literary and imaginative. Combining a horse with a horn to produce the impossible fantasy blend of a unicorn is a popular example of the literary imagination at work. It may seem that this process of combination must be secondary and parasitic: surely stable and integrated concepts (horse, horn) and small spatial stories (horses run, horns impale) must be in place and must have arisen by elementary processes of perception and memory before second-order processes like integration and blending can work on them.
Only very recently—in the last few years—has neuroscience begun to suggest that the opposite might be true. It would be a mistake to hang too much at this stage on the specific details of the various neuroscientific theories that have arisen, but a general principle is emerging, and it is this: At the most basic levels of perception, of understanding, and of memory, blending is fundamental.
When we perceive or remember or think about a particular horse or horse in general, the horse seems to us whole. At the seemingly simple level of vision, a horse just looks like a visual whole. It is whole in our sight. Our recognition of it seems whole, not piecemeal or fragmented.
We expect our neurobiology to work at least loosely the way our perception seems to work, and we think (wrongly) that introspection reveals at least roughly how our perception works. We expect our phenomenology to indicate the nature of neurobiology. But it does not. It appears that there may be no anatomical site in the brain where a perception of a horse or a concept horse resides, and, even more interestingly, no point where the parts of the perception or concept are anatomically brought together. The horse looks to us like obviously one thing; yet our visual perception of it is entirely fragmented across the brain. What the brain does is not at all what we might have expected. The visual perceptions of color, texture, movement, form, topological attributes, part-whole structure, and so on occur in a fragmentary fashion throughout the brain and are not assembled in any one place. This is a surprise, like the surprise of learning that although the visual field is projected upside-down onto the retina because of the simple optics of a lens, there is no place in the brain where that image is reassembled and turned right-side-up. The qualities we believe our perception to have are in many ways not at all the qualities that neuroscientists are finding in the neurobiological activity that underlies perception.
As Gerald Edelman writes, “Objects and many of their properties are perceived as having a unitary appearance; yet these unitary perceptions are the consequences of parallel activity in the brain of many different maps, each with different degrees of functional segregation. Many examples could be cited; a striking case is the extra-striate visual cortex, with its different areas mediating color, motion, and form, each in different ways.”
How, then, do these fragments end up seeming to us like one perception or one concept—one horse? This is known in neuroscience as the “binding problem.” The binding problem is part of a more general problem, integration. For example, given that form, color, and motion are all processed differently and at different places in the brain, and that no one place in the brain receives projections from all the brain sites involved in this processing, how is it that the form, color, and motion of the horse are integrated?
Visual integration, although perhaps the archetype of integration, is only one aspect of integration. The concept of the horse, even the perception of a horse, may integrate the sound the horse makes, the way it feels to sit on a horse’s back, the way it feels to ride a horse, the dynamic image schema of mounting the horse, the convexity of the horse’s torso, and the sound its swishing tail makes. Nor are even these elements unitary—they also each require integration. The part-whole relational structure of the horse, its overall form, and its coloring, for example, which seem to us so unitary, also require integration.
The horse that seems one thing corresponds to a widely distributed fragmentation in the brain. Mentally, the unitary horse is a fabulous blend.
Neuroscientific attempts to solve the general problem of perceptual and conceptual integration are speculative and interesting. All of them depend upon models of integration through close timing of related neuronal events, rather than gathering of information in an anatomical site. Antonio Damasio, for example, has proposed a model of “convergence” according to which the brain contains records of the combinatorial relations of fragmentary records; the recall of entities or events arises from a reactivation, very tightly bound in time, of fragmentary records contained in multiple sensory and motor regions. Mental evocations that seem to us so unitary and solid are instead always fleeting reblendings of reactivated fragments in a very tight and intricate interval of time.
Damasio sees his speculative proposal for explaining “convergence” as generally compatible with the theory of Neuronal Group Selection proposed by Gerald Edelman and his associates, who have made an attempt to solve the binding problem (and more generally the problem of integration) by invoking a conjectural neurobiological process called “reentrant signaling.” Differing in detail from Damasio’s “convergence” model, reentrant signaling nonetheless also depends upon the coordination and integration of distributed fragmentary operations. In any chunk of time, reentry “involves parallel sampling from a geometric range of spatially extended maps made up of neuronal groups.” Vision and perceptual categorization supply Edelman’s most thoroughly worked examples. During response to the visual stimulus world, reentrant signaling
acts to coordinate inputs and resolve conflicts between the responses of different functionally segregated maps. Visual reentry allows each mapped region to use discriminations made by other regions (about borders, movement, etc.) for its own operations. This process allows the various responses to aspects of the stimulus world to remain segregated and distributed among multiple brain areas and still constitute a unified representation. Reentrant integration obviates the need for a higher-level command center, or a “sketch.”
Edelman generalizes this specific model of reentry for visual areas to “include functional correlations important to concept formation, consciousness, and speech.”
For our purposes here, the details of these proposals are not at issue. We must be prepared to see many future attempts to solve the problem of integration. Nor, as the study of the mind and the study of the brain themselves attempt to blend and integrate, should we leap to facile connections, making assumptions, for example, that the conceptual blending we see in the tale of the ox and the donkey must somehow be explained by simple models of integration in perception and recall.
But there is a general principle that may help to connect the study of tne brain with the study of the mind: Blending is a basic process; meaning does not reside in one site but is typically a dynamic and variable pattern of connection over many elements. Our conscious experience seems to tell us that meanings are whole, localized, and unitary. But this is wrong. Blending is already involved in our most unitary and literal perception and conception of basic physical objects, such as horse and horn, and in our most unitary and literal perception and conception of small spatial stories, such as horse moves and horn impales.
When we pay close attention, most mental events appear to involve blending of one sort or another. Whenever we see something as something—when we look at the street and see a woman getting into a car—we are blending our sensory experience with abstract conceptual structure. The sensory perception and the abstract conceptual structure do not look as if they are partitioned into two different components. The perception of someone getting into a car does not seem fragmentary, with one part corresponding to the visual experience and another part corresponding to the specific action and its status as an instance of a general action. These two very different things—the sensory activity and the conceptual activity—do not seem at all like separate parcels. When they do, something has gone wrong. Perception and conception seem to us one unitary whole, but they involve blending.
Anytime we perceive something that we take to be part of a larger whole (part of a figure, part of an event, part of a small story, part of a melody, and so on), we are blending perceptual experience with the recall of that whole. Whenever we categorize new information, we are blending the new information and the established category.
When we recognize what looks like a simple event of force, we must blend together very abstract image schemas of force dynamics with parts of the specific perception. Blending abstract image schemas together is a basic conceptual operation; blending abstract image schemas with specific perceptual experience is a basic conceptual operation. When we see a car go through an intersection, for example, we must blend very many things, including the image schema of path, the perception corresponding to the car’s movement, the image schema of container, and the perception corresponding to the bounded area of the intersection.
Temporality seems to be as dependent upon blending as is spatiality. Even a simple mental event like looking at a street and remembering the red car that went down it yesterday depends upon an impossible blend: today’s perceptual experience of the street and recall of yesterday’s perceptual experience of the street. This impossible blending of realities that belong to different temporal spaces is a routine part of understanding. Slightly highlighted, we can notice the impossibility and the blending, as when we are asked to blend temporal spaces to produce the “race” between Great America II and Northern Light.
There is one transcendent story of the mind that has appeared many times in many avatars. In its essential lines, it claims that there are certain basic, sober, and literal things the mind does; that imaginative and literary acts are parasitic, secondary, peripheral, exotic, or deviant; and that when neuroscience gets its act together, we will come to understand that the brain does things pretty much in the ways we always expected. On this logic, since imaginative and literary acts are peripheral and exotic, they can safely be ignored while, as serious scientists, we investigate the basics.
This story, which is itself just an imaginative story, has been the pretext for offering indefinitely many “first pass” scientific models of some of our supposedly basic mental operations. These models ignore what appear to be more sophisticated and exotic mental events, like blending, on the claim—usually taken for granted—that first we must explain a few basic operations, and then we can work on explaining the more imaginative operations that are parasitic upon the basic mind.
It is possible that this story is just wrong at its core. The brain does not seem to work at all in the ways we expected it to, based on our notion of stable and unitary concepts. On the contrary, our notion of concepts as stable and unitary seems to be a false guide to neurobiology. Blending may seem exotic to us, but in fact it may have a fundamental neurobiological analogue. It should not be surprising if blending turns out to be basic, not exotic, in the everyday mind. Certainly there is considerable evidence that blending is a mainstay of early childhood thought.
A two-year-old child who is leading a balloon around on a string may say, pointing to the balloon, “This is my imagination dog.” When asked how tall it is, she says, “This high,” holding her hand slightly higher than the top of the balloon. “These,” she says, pointing at two spots just above the balloon, “are its ears.” This is a complicated blend of attributes shared by a dog on a leash and a balloon on a string. It is dynamic, temporary, constructed for local purposes, formed on the basis of image schemas, and extraordinarily impressive. It is also just what two-year-old children do all day long. True, we relegate it to the realm of fantasy because it is an impossible blended space, but such spaces seem to be indispensable to thought generally and to be sites of the construction of meanings that bear on what we take to be reality.
A scientific model of thought frequently tries to begin with what is basic, on the claim that scientists must do first things first and second things second and exotic things sometime next century. This theoretical vehicle for getting an explanation of mind off the ground has crashed and burned often. It is not implausible that the concepts behind such models are wrong, that something like imaginative blending and integration are basic, and that an explanation that cannot handle “This is my imagination dog” has no hope of ever getting to even the most basic perceptions and meanings, like the perception of a dog or the meaning of “A dog has four legs,” or even what is involved in an infant’s pointing at a dog and saying “Doggie!”
In the hard sciences, which attract so much emulation and envy, unusual events are not dismissed as peripheral. On the contrary, unusual events often command the most attention, on the principle that they are the most likely to reveal general processes. The usual case can be accounted for as the result of general processes working in minimal particular conditions. A scientific experiment that tests for fundamental mechanisms is often elaborate and weird, something not likely to occur on its own (particle accelerators, odd refractions of light involving specially engineered mirrors, chemical reactions in zero gravity, and so on). Physicists who noticed that the orbit of Mercury did not quite follow Newtonian theory did not ignore it as an exotic event. On the contrary, it became the central event, calling for new theories and extraordinary new experiments. One of these experiments, for example, involved waiting for a total eclipse of the sun, in order to measure the positions of stars around the outer edge of the eclipse, so as to compare these measurements with theoretically predicted positions, to see whether the light from the stars was being bent by the mass of the sun. A more unusual, exotic, and “marginal” astrophysical event can hardly be imagined, but such events often constitute the touchstones of the hard sciences. Physicists interested in the fundamental laws of the universe had to voyage to the tropics to get the most revealing data. Cognitive scientists might interpret this as a parable for their research. The hard sciences do not lump apparendy odd events into the category of what we don’t need to explain, but rather give them special attention. It is not clear that someone in the cognitive sciences who hopes her discipline will attain to the prestige of the hard sciences should behave any differently. In particular, the methods of the hard sciences give no legitimacy to the story that we can tell in advance what is basic and what is exotic and form our theories by ignoring what we think is exotic.
The processes of the literary mind are usually considered to be different from and secondary to the processes of the everyday mind. On that assumption, the everyday mind—with its stable concepts and literal reasoning—provides the beginnings for the (optional) literary mind. On the contrary, processes that we have always considered to be literary are at the foundation of the everyday mind. Literary processes like blending make the everyday mind possible.