Scholars have long contemplated the connection between language and thought—and to what degree the two are intertwined—by asking whether language is somehow an essential prerequisite for thinking.
British philosopher and mathematician Bertrand Russell answered the question with a flat yes, asserting that language’s very purpose is “to make possible thoughts which could not exist without it.” But even a cursory glance around the natural world suggests why Russell may be wrong: No words are needed for animals to perform all sorts of problem-solving challenges that demonstrate high-level cognition. Chimps can outplay humans in a strategy game, and New Caledonian Crows make their own tools that enable them to capture prey.
Still, humans perform cognitive tasks at a level of sophistication not seen in chimps—we can solve differential equations or compose majestic symphonies. Is language needed in some form for these species-specific achievements? Do we require words or syntax as scaffolding to construct the things we think about? Or do the brain’s cognitive regions devise fully baked thoughts that we then convey using words as a medium of communication?
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
Evelina Fedorenko, a neuroscientist who studies language at the McGovern Institute for Brain Research at the Massachusetts Institute of Technology, has spent many years trying to answer these questions. She remembers being a Harvard University undergraduate in the early 2000s, when the language-begets-thought hypothesis was still highly prominent in academia. She herself became a believer.
When Fedorenko began her research 15 years ago, a time when new brain-imaging techniques had become widely available, she wanted to evaluate this idea with the requisite rigor. She recently co-authored a perspective article in Nature that includes a summary of her findings over the ensuing years. It makes clear that the jury is no longer out, in Fedorenko’s view: language and thought are, in fact, distinct entities that the brain processes separately. The highest levels of cognition—from novel problem-solving to social reasoning—can proceed without an assist from words or linguistic structures.
Language works a little like telepathy in allowing us to communicate our thoughts to others and to pass to the next generation the knowledge and skills essential for our hypersocial species to flourish. But at the same time, a person with aphasia, who are sometimes unable to utter a single word, can still engage in an array of cognitive tasks fundamental to thought. Scientific American talked to Fedorenko about the language-thought divide and the prospects of artificial intelligence tools such as large language models for continuing to explore interactions between thinking and speaking.
[An edited transcript of the interview follows.]
How did you decide to ask the question of whether language and thought are separate entities?
Honestly, I had a very strong intuition that language is pretty critical to complex thought. In the early 2000s I really was drawn to the hypothesis that maybe humans have some special machinery that is especially well suited for computing hierarchical structures.And language is a prime example of a system based on hierarchical structures: words combine into phrases and phrases combine into sentences.
And a lot of complex thought is based on hierarchical structures. So I thought, ‘Well, I’m going to go and find this brain region that processes hierarchical structures of language.’ There had been a few claims at the time that some parts of the left frontal cortex are that structure.
But a lot of the methods that people were using to examine overlap in the brain between language and other domains weren’t that great. And so I thought I would do it better. And then, as often happens in science, things just don’t work the way you imagine they might. I searched for evidence for such a brain region—and it doesn’t exist.
You find this very clear separation between brain regions that compute hierarchical structures in language and brain regions that help you do the same kind of thing in math or music. A lot of science starts out with some hypotheses that are often based on intuitions or on prior beliefs.
My original training was in the [tradition of linguist Noam Chomsky], where the dogma has always been that we use language for thinking: to think is why language evolved in our species. And so this is the expectation I had from that training. But you just learn, when you do science, that most of the time you’re wrong—and that’s great because we learn how things actually work in reality.
What evidence did you find that thought and language are separate systems?
The evidence comes from two separate methods. One is basically a very old method that scientists have been using for centuries: looking at deficits in different abilities—for instance, in people with brain damage.
Using this approach, we can look at individuals who have impairments in language—some form of aphasia. Aphasia has been studied as a condition for centuries. For the question of how language relates to systems of thought, the most informative cases are cases of really severe impairments, so-called global aphasia, where individuals basically lose completely their ability to understand and produce language as a result of massive damage to the left hemisphere of the brain. You can ask whether people who have these severe language impairments can perform tasks that require thinking. You can ask them to solve some math problems or to perform a social reasoning test, and all of the instructions, of course, have to be nonverbal because they can’t understand linguistic information anymore. Scientists have a lot of experience working with populations that don’t have language—studying preverbal infants or studying nonhuman animal species. So it’s definitely possible to convey instructions in a way that’s nonverbal. And the key finding from this line of work is that there are people with severe language impairments who nonetheless seem totally fine on all cognitive tasks that we’ve tested them on so far.
There are individuals who have been now tested on many, many different kinds of tasks, including tasks that involve what you may call thinking, such as solving math problems or logic puzzles or reasoning about what somebody else believes or reasoning about the physical world. So that’s one big chunk of evidence from these populations of people with aphasia.
What is the other method?
A nicely complementary approach, which started in the 1980s and 1990s, is a brain-imaging approach. We can measure blood flow changes when people engage in different tasks and ask questions about whether the two systems are distinct or overlapping—for example, whether your language regions overlap with regions that help you solve math problems. These brain-imaging tools are really good for these questions. But before I could ask these questions, I needed a way to robustly and reliably identify language areas in individual brains, so I spent the first bunch of years of my career developing tools to do this.
And once we have a way of finding these language regions, and we know that these are the regions that, when damaged in adulthood, lead to conditions such as aphasia, we can then ask whether these language regions are active when people engage in various thinking tasks. So you can come into the lab, and I can put you in the scanner, find your language regions by asking you to perform a short task that takes a few minutes—and then I can ask you to do some logic puzzles or sudoku or some complex working memory tasks or planning and decision-making. And then I can ask whether the regions that we know process language are working when you’re engaging in these other kinds of tasks. There are now dozens of studies that we’ve done looking at all sorts of nonlinguistic inputs and tasks, including many thinking tasks. We find time and again that the language regions are basically silent when people engage in these thinking activities.
So what is the role of language, if not for thinking?
What I’m doing right now is sharing some knowledge that I have that you may have only had a partial version of—and once I transmit it to you through language, you can update your knowledge and have that in your mind as well. So it’s basically like a shortcut for telepathy. We can’t read each other’s mind. But we can use this tool called language, which is a flexible way to communicate our inner states, to transmit information to each other.
And in fact, most of the things that you probably learned about the world, you learned through language and not through direct experience with the world. So language is very useful. You can easily imagine how it would confer evolutionary advantages: by facilitating cooperative activities, transmitting knowledge about how to build tools and conveying social knowledge. As people started living in larger groups, it became more important to keep track of various social relationships. For example, I can tell you, “Oh, I don’t trust that guy.” Also, it’s very hard to transmit knowledge to future generations, and language allows us to do that very effectively.
In line with the idea that we have language to communicate, there is accumulating evidence from the past few decades that shows that various properties that human languages have—there are about 7,000 of them spoken and signed across the world—are optimized for efficiently transmitting information, making things easy to perceive, easy to understand, easy to produce and easy to learn for kids.
Is language what makes humans special?
We know from brain evolution that many parts of the cortical sheet [the outer layer of the brain] expanded a lot in humans. These parts of the brain contain several distinct functional systems. Language is one of them. But there’s also a system that allows us to reason about other minds. There’s a system that supports novel problem-solving. There’s a system that allows us to integrate information across extended contexts in time—for example, chaining a few events together. It’s most likely that what makes us human is not one “golden ticket,” as some call it. It’s not one thing that happened; it’s more likely that a whole bunch of systems got more sophisticated, taking up larger chunks of cortex and allowing for more complex thoughts and behaviors.
Do the language and thinking systems interact with each other?
There aren’t great tools in neuroscience to study intersystem interactions between language and thought. But there are interesting new opportunities that are opening up with advances in AI where we now have a model system to study language, which is in the form of these large language models such as GPT-2 and its successors. These models do language really well, producing perfectly grammatical and meaningful sentences. They’re not so good at thinking, which is nicely aligning with the idea that the language system by itself is not what makes you think.
But we and many other groups are doing work in which we take some version of an artificial neural network language model as a model of the human language system. And then we try to connect it to some system that is more like what we think human systems of thought look like—for example, a symbolic problem-solving system such as a math app. With these artificial intelligence tools, we can at least ask, “What are the ways in which a system of thought, a system of reasoning, can interact with a system that stores and uses linguistic representations?” These so-called neurosymbolic approaches provide an exciting opportunity to start tackling these questions.
So what do large language models do to help us understand the neuroscience of how language works?
They’re basically the first model organism for researchers studying the neuroscience of language. They are not a biological organism, but until these models came about, we just didn’t have anything other than the human brain that does language. And so what’s happening is incredibly exciting. You can do stuff on models that you can’t do on actual biological systems that you’re trying to understand. There are many, many questions that we can now ask that had been totally out of reach: for example, questions about development.
In humans, of course, you cannot manipulate linguistic input that children get. You cannot deprive kids of language, or restrict their input in some way, and see how they develop. But you can build these models that are trained on only particular kinds of linguistic input or are trained on speech inputs as opposed to textual inputs. And then you can see whether models trained in particular ways better recapitulate what we see in humans with respect to their linguistic behavior or brain responses to language.
So just as neuroscientists have long used a mouse or a macaque as a model organism, we can now use these in silico models, which are not biological but very powerful in their own way, to try to understand some aspects of how language develops or is processed or decays in aging or whatnot.
We have a lot more access to these models’ internals. The methods we have for messing with the brain, at least with the human brain, are much more limited compared with what we can do with these models.