When you look closely at how we use language, you find that a lot of what we say is metaphorical—we talk about certain things as though they were other things. We describe political campaigns as horse races: “Senator Jones has pulled ahead.” Morality is cleanliness: “That was a dirty trick.” And understanding is seeing: “New finding illuminates the structure of the universe.”
People have known about metaphor for a very long time. Until the end of the 20th century, most everyone agreed on one particular explanation, neatly articulated by Aristotle and carried down through the centuries. Metaphor was seen as a strictly linguistic device—a kind of catchy turn of phrase—in which you call one thing by the name of another thing that it's similar to. This is probably the definition of metaphor you learned in high school English. On this view, you can metaphorically say that “Juliet is the sun” if and only if Juliet and the sun are similar—for instance, if they are both particularly luminous.
But in their 1980 the book Metaphors We Live By, George Lakoff and Mark Johnson proposed an explanation for metaphorical language that flew in the face of this received wisdom. They reasoned that if metaphor is just a free-floating linguistic device based on similarity, then you should be able to metaphorically describe anything in terms of anything else that it's similar to. But Lakoff and Johnson observed that real metaphorical language as actually used isn't haphazard at all. Instead, it's systematic and coherent.
It's systematic in that you don't just metaphorically describe anything as anything else. Instead, it's mostly abstract things that you describe in terms of concrete things. Morality is more abstract than cleanliness. Understanding is more abstract than seeing. And you can't reverse the metaphors. While you can say “He's clean” to mean he has no criminal record, you can't say “He's moral” to mean that he bathed recently. Metaphor is unidirectional, from concrete to abstract.
Metaphorical expressions are also coherent with one another. Take the example of understanding and seeing. There are lots of relevant metaphorical expressions, for example “I see what you mean,” and “Let's shed some light on the issue,” and “Put his idea under a microscope and see if it actually makes sense.” And so on. While these are totally different metaphorical expressions—they use completely different words—they all coherently cast certain aspects of understanding in terms of specific aspects of seeing. You always describe the understander as the seer, the understood idea as the seen object, the act of understanding as seeing, the understandability of the idea as the visibility of the object, and so on. In other words, the aspects of seeing you use to talk about aspects of understanding stand in a fixed mapping to one another.
These observations led Lakoff and Johnson to propose that there was something going on with metaphor that was deeper than just the words. They argued that the metaphorical expressions in language are really only surface phenomena, organized and generated by mappings in people's minds. For them, the reason metaphorical language exists and the reason why it's systematic and coherent is that people think metaphorically. You don't just talk about understanding as seeing; you think about understanding as seeing. You don't just talk about morality as cleanliness; you think about morality as cleanliness. And it's because you think metaphorically—because you systematically map certain concepts onto others in your mind—that you talk metaphorically. The metaphorical expressions are merely the visible tip of the iceberg.
As explanations go, this one covers all the bases. It's elegant in that it explains messy and complicated phenomena (the various metaphorical expressions we have that describe understanding as seeing, for instance) in terms of something much simpler—a structured mapping between the two conceptual domains in people's minds. It's powerful in that it explains things other than metaphorical language—recent work in cognitive psychology shows that people think metaphorically even in the absence of metaphorical language; affection as warmth, morality as cleanliness. As a result, the conceptual metaphor explanation helps to explain how it is that we understand abstract concepts like affection or morality at all—by metaphorically mapping them onto more concrete ones. In terms of utility, the conceptual metaphor explanation has generated extensive research in a variety of fields; linguists have documented the richness of metaphorical language and explored its diversity across the globe, psychologists have tested its predictions in human behavior, and neuroscientists have searched the brain for its physical underpinnings. And finally, the conceptual metaphor explanation is transformative—it flies in the face of the accepted idea that metaphor is just a linguistic device based on similarity. In an instant, it made us rethink 2000 years of received wisdom.
This isn't to say that the conceptual metaphor explanation doesn't have its weaknesses, or that it's the final word in the study of metaphor. But it's an explanation that casts a huge shadow. So to speak.