Louder Than Words

When I first read about this ingenious behavior, I found it fascinating.² Does the bear have the mental flexibility to envision what it looks like to others and the creativity to figure out how to conceal itself? Or is this nose covering just a trick that evolution has dropped into the polar bear's quiver of built-in behaviors—a freak behavior that happened to confer a survival advantage and was therefore selected for over the course of millennia?

Now, although there's doubtless a lot more to say about these charismatic megafauna, this is not a book about polar bears. It's a book about you and, more specifically, how you understand language. So consider, if you will, what you did when you opened this book and started reading the first paragraph. You cast your eyes over the letters that made up the words. You recognized familiar words like bear and seal and hunting and snow. That all seems pretty straightforward—it's the kind of thing a well-written piece of software or well-trained parrot could do. But then you started doing things that were a little deeper. Once you knew what the words were, you began to find meaning in them. You knew what type of animals and objects the nouns referred to and what types of actions and events the verbs described. But you didn't stop at the words. You made sense of the sentences they made up, sentences that I'm almost certain you had never encountered before (unless this not your first time reading this book). And the things the sentences described probably came to life—the bear scooting along its belly through the snow and the ingenious but awkward way it would have to hold its paw over its nose. Maybe you even went so far as to virtually "see" the arctic scene in your mind's eye.

And then—and here's the really remarkable part—you went way beyond that. You filled in details that were never explicitly mentioned. How do I know? You see, polar bears, as you surely surmised, hide their dark muzzles because the thick fur that covers their bodies, including their paws but not including their noses, is white. And they live surrounded by snow and ice, which for the most part is also white. But here's the thing. I actually never mentioned anything about color. If you look back at the first paragraph of this chapter, you'll see that the whiteness of the snow and of the bear and the blackness of its nose are completely implied. You colored in the picture. And it's a good thing you did, because without color, the story makes absolutely no sense at all. There's no other obvious reason for a polar bear to cover its nose.

How do you manage to do all this? How do you take scribbles on a page or for that matter the pops, buzzes, and hums of human speech and make them mean something to you? How do you know what the words and sentences mean and how do you fill in the gaps? How do you do what you're doing right now? That's the mystery of meaning. And that is in fact what this book is about.

Meaning is not only constant; it's also critical. We use language to make sense of the world. We use it almost any time we interact with other people: to flirt, command, inform, beg, and form social bonds. A few words can change our minds, change our marital status, or change our religion. Words affect who we are. As a species, language is our most powerful and pervasive tool. With language, we can communicate what we think and who we are. Without language, we would be isolated. We would have no fiction, no history, and no science. To understand how meaning works, then, is to understand part of what it is to be human.

And not just human, but uniquely human. No other animal can do what we can with language. Of course, parts of human language have homologues in other animals. People talk fast, and sentences can be extremely complicated, but zebra finches sing tunes that rival our speed and complexity. Humans can drone on and on, but even a filibustering senator doesn't outlast humpback whales, whose songs can continue for hours. And although the human ability to combine words in new ways seems pretty unique, it's seen on a more limited scale in bees, who dance messages to each other that combine information about the orientation, quality, and distance of food sources. What's special about human language—what marks it as distinct from every other naturally occurring form of communication in the known universe—is that we can use it to convey pretty much any meaning that we want. A bee can waggle its abdomen until it falls off, but it will never communicate anything beyond what it's programmed to—it can't say that the weather's likely to clear up, that it had a decent night's sleep, or that it's looking forward to the weekend because it has a hot date with a hydrangea. Human language, in contrast to all other animal communication systems, is open-ended. We can talk about things that exist, like inarticulate presidential candidates and rail-thin models, or even things that don't, like Martian anthropologists or vegetarian zombies. And, for the most part, other people—at least people who speak our language and have normally functioning cognitive systems—are able to understand us. No other animal can do this. And because this level of meaning making is unique to our species, determining how it works brings us one step closer to knowing what distinguishes us from other animals.

There are other, more practical reasons to pursue the science of meaning. Imagine computer systems that truly understand you when you talk to them (Siri or Watson on steroids) or that can automatically translate from one human language to another. No reasonable Star Trek–worthy future would be complete without them. Understanding how meaning works can also help us improve the way we teach foreign languages. And it can lead to restorative therapies and technologies for people who have suffered brain damage that impairs their ability to understand or produce language meaningfully.

For all these reasons, language has held a privileged spot in science and philosophy throughout history. For centuries, philosophers have asked what is it that we humans have that our tongue-tied relatives don't; what cognitive capacities evolution has endowed us with that allow us to understand—and appreciate—sonnets and songs, exhortations and explanations, newspapers, and novels. And there are half a dozen academic disciplines dedicated to different aspects of language: from English and foreign languages to communications, semantics, psycholinguistics, cognitive linguistics, and neurolinguistics. Thanks to research in these fields, we now know a lot about the grammar of sentences, about how people articulate speech, and how to best teach a foreign language.

But for the most part, we've failed to answer the most important question of all. Language matters to us because it is a vehicle for meaning—it allows us to take the desires, intentions, and experiences in our heads and transmit a signal through space that makes those thoughts pop up in someone else's head. We don't study French in order to form perfectly grammatical French sentences; we learn it to communicate. We don't read fiction because the words look appealing on the page but because of the transporting flood of sights, sounds, places, and ideas that good writing evokes. And yet, almost no one, from lay people to linguists, really knows how meaning works.

That is, until recently. This is the age of cognitive science. Had we been born earlier, we might be exploring new continents. Born later, we might be gallivanting through the stars. But right now, at this time in our history, the vast, tantalizing expanse that begs to be discovered is the human mind. And some cognitive scientists, like me, have started to turn their attention to meaning. Over the past decade, a few key experimental advances have quickly elevated meaning to "hot topic" status in cognitive science. Using fine measures of reaction time, eye gaze, and hand movement, as well as brain imaging and other state-of-the-art tools, we've started to scrutinize humans in the act of communicating. We can now peer inside the mind and thereby put meaning in its rightful place at the center of the study of language and the mind. With these new tools, we've managed to catch a glimpse of meaning in action, and the result is revolutionary. The way meaning works is much richer, more complex, and more personal than we ever would have predicted.

This book tells the story of what we've discovered so far.

However, even in the absence of solid empirical evidence, theories about how meaning works have developed and thrived. Over the years, most linguists, philosophers, and cognitive psychologists have come to settle on a particular story, which probably isn't so different from your intuitive sense of meaning. When you contemplate meaning in your daily life, it's likely because you're wondering (or perhaps arguing about) what a given word means. It might be a word in your own language: What does obdurate mean? (Stubbornly persistent in wrongdoing, in case you were wondering.) How about necrophagia? (Eating the dead.) Or epicaricacy? (Taking pleasure at others' misfortune.) Or it could be a word in another language: What does the formidable German word Geschwindigkeitsbegrenzung mean? (Speed limit.) In general, you're probably most aware of meaning when you're thinking about definitions. This is also the starting point for the traditional theory of meaning: words have meanings that are like definitions in your mind.

What would it be like if meaning worked this way? When you think about it, a definitional meaning would need to have two distinct parts. The first is the definition itself. This is a description of what the word means. It's articulated in a particular language, like English, and is supposed to be a usable characterization of the meaning. But there's a second part, too, which is implicit. The definition characterizes something in the world. So speed limit (or if you prefer, Geschwindigkeitsbegrenzung ) actually refers to something that exists in real life, independent of your knowledge about it—whether you know that there's a speed limit, or what it is, you can still get pulled over for driving faster than the number on the sign. So both the mental definition and the actual thing in the world that the word refers to are each critical parts of the meaning of a word.

Many philosophers have taken it as a given that these two parts are all you need to characterize meaning.³ And they've gone on to argue for centuries about which of the two parts is more important—the mental definition or the real world. But the important question for our purposes—to understand how people understand—is to ask how a definitional theory of meaning like this could explain the things we do with language. Do we really have these definitions in our minds? If so, where do they come from? How could we use them to plan a sequence of words? How could we use them to understand something that someone else has said?

This is where things get a little more complicated. As with any definition, your mental definitions would presumably need to be articulated in some language. But what language? Your first thought might be that it should be your native language, so English words have mental definitions in English, and German words have mental definitions in German. Except, when you follow that idea to its logical conclusion, there's a problem. If English words are defined in your mind in terms of other English words, then how do you understand the definitions themselves? You end up going in circles. Here's an illustration of the problem from a real-life situation that you might be able to relate to.⁴ Suppose you don't speak Japanese. But you're at a train station in Tokyo, and there's a sign you want to look up the meaning of. So you pull out your dictionary and look up the characters, but at that point you realize, to your chagrin, that instead of a Japanese-English bilingual dictionary, you accidentally bought a Japanese-Japanese monolingual dictionary. Oops. On the sign, there's a squiggly character with a horizontal line and some dots, so you look that up in your dictionary, but, regrettably, the definition is nothing but a long string of many more characters that you also don't recognize. You could try to look these up, in turn, but you'd just get more of the same. The problem is the same one that you would have if your mental definitions were expressed in your native language. Definitions expressed in a particular language don't mean anything unless you understand the language already. So in understanding the word polar bear, say, it wouldn't work to go through a process of activating an English definition of polar bear (a large, white, carnivorous bear that inhabits arctic regions) in your mind. This definition wouldn't be any more meaningful than the polar bear that you started out with.

One solution to this problem is to suppose that we have some other system in our mind—some way to encode ideas and thoughts and reasoning that doesn't use English or any real language. This mental language would need to have a lot of the stuff that a real language has—it would still have to be able to refer to things in the world, as well as properties, relations, actions, events, and so on—anything that we can think about and understand language about. In other words, we might be thinking using something like a language of thought or Mentalese .⁵ Simply stated, the language of thought hypothesis is that the meanings of words and sentences in any real language are articulated in people's minds in terms of this other, mental language. Mentalese is supposed to be like a real language in that there are words that mean things and can combine with one another, but, unlike a real language, it doesn't sound like anything or look like anything. So, in Mentalese, we have a word that represents speed limits, and another for epicaricacy, and another for polar bears, and so on. To understand a real language like English or Chinese, we need to translate the words we hear or read into Mentalese. So the language of thought hypothesis breaks mental definitions out of their self-referential circle by seeing the human capacity for meaning as akin to using a bilingual dictionary instead of a monolingual one. If you showed up at a Japanese train station with a Japanese-English dictionary, you could understand what the Japanese characters meant by looking them up in the dictionary, because the dictionary translates them into words a language you already know. And by analogy, the language of thought hypothesis states that for each word that we know, we have a mental entry that includes a definition articulated in Mentalese. This is one of the most important and influential ideas people have had about meaning and the mind.

But even if Mentalese gets us out of the vicious circle of words defined in terms of other words, it still only gets us part way to meaning. That's because it doesn't deal with the other half of a definitional theory of meaning—the things in the world that the Mentalese words refer to. According to the language of thought hypothesis, the words of Mentalese are related to the world through a symbolic relationship. For instance, when you read the words polar bear and translate them into whatever your Mentalese word for polar bear is, let's call it 9us&'~ (as a reminder that it's not supposed to be pronounceable), that word has meaning by dint of the set of things in the world that are actually polar bears. So a sentence like The polar bear mostly blends in with its icy, snowy surroundings has meaning because it describes a situation in the world where a thing appropriately designated by your symbol for polar bear is in fact doing something designated by your symbol for blends into something designated by your symbols for icy, snowy surroundings.

Over the centuries, this has come to be the leading idea about how meaning works. Words are meaningful because you have mental definitions for them—articulated in Mentalese—that match up to things in the real world.

One way to think about this issue is using a version of a thought experiment known as the Chinese Room argument.⁶ Say you're sitting in an enclosed room with two slots in it. Occasionally, someone will slide a card written in Chinese characters into the room through one of the slots. Now, you don't know any Chinese, but your job is to look these characters up in a book. The book will have some other characters next to the one you looked up, and you're supposed to find a card with those other characters on them and slide it out of the room through the other slot. Because you don't know Chinese, you have no idea what's on the cards, but people outside the room, who do know Chinese, think that the person in the room must certainly be a native Chinese speaker because the responses that come out of the room are perfectly appropriate rejoinders to the messages that they slip into the room. Of course, this is only possible if the book you're looking up the answers in is really well designed. But the question is: Do you understand Chinese? I suspect you'll agree that, no, of course you don't. We can apply the same reasoning to the language of thought hypothesis as an explanation of how meaning works. The Chinese characters in this example are like the words of Mentalese. Simply identifying and arranging symbols in some language, even if those symbols represent something in the real world, isn't enough to make meaning. It's not enough to say you've understood something.

This is one of the big problems with the language of thought hypothesis. And when you start to apply a little pressure, other cracks start to appear. For one, where does Mentalese come from? If it's something that's learned, then it certainly can't be learned through one's native language, because that creates another vicious cycle: How could we learn Mentalese based on English if we only understand English through Mentalese? So if Mentalese can't be learned from language, then that means that—if there is such a thing as Mentalese—it has to exist in our minds before we even start to learn language. In other words, in order to learn the English polar bear, we have to already have a Mentalese symbol representing polar bears. And this also means that people who speak different languages must all have the same underlying concepts—a polar bear is a polar bear is a polar bear. There's good reason to question all of these claims.

Even the greatest strength of the language of thought hypothesis—the simplicity of Mentalese symbols—is gained at substantial cost. The idea that the weight of meaning might be carried by Mentalese symbols is quite powerful and appealing, because those symbols would be so simple. Symbols are pointers that just tell you what things in the world they refer to. To understand what the English word polar bear means is to have a symbol 9us&'~ that refers to actual polar bears in the world. To understand what the English word dog