This Is Your Brain on Metaphor
Neuroscientists use fMRI to look at our most treasured turns of phrase
Hi, this is Steven Cherry for IEEE Spectrum’s “Techwise Conversations.”
We’ve done several bioengineering shows in the past year, and this is another one, about functional magnetic resonance imaging, or fMRI. But first, a bit of autobiography. I have a personal life of sorts, in which I’m also a writer and a writing instructor. My personal e-mail address is firstname.lastname@example.org, which I snatched up in the early days of Gmail’s beta test, and my personal blog at WordPress is called Metaphorical.
So it’s fair to say the subject of metaphor is pretty close to my heart.
A couple of weeks ago we had a show with a neurosurgeon who’s been busy putting devices into people’s brains. The great thing about both fMRI and brain implant studies is that they help us understand the puzzling duality that Descartes first wrote about 350 years ago: What’s going on in the brain when people are busy thinking?
I find this sort of research to be incredibly interesting but also frustrating, because I want to ask, But what’s going on with metaphors? The problem is this: We expect the corresponding brain activity to involve the more linguistic centers of the brain, but to the extent that metaphors ground abstract ideas in more easily depicted ones, such as physical objects or spatial distances, we would expect to see activity in very different, more sensory-related areas of the brain.
So when Elizabeth Barrett Browning wrote that she could love to the depth and breadth and height her soul could reach, was the brain firing in its linguistic areas, in the romantic love areas, or in the sensory and spatial ones?
There really needs to be someone studying this, and as it turns out, there is. Dr. Krishnankutty Sathian—or Krish, as his colleagues and friends call him—is a professor of neurology, rehab medicine, and psychology, and he’s a coauthor of a new study with the title, “Metaphorically Feeling: Comprehending Textural Metaphors Activates Somatosensory Cortex,” published last month in the journal Brain & Language.
He joins us today by phone from his office at Emory University, in Atlanta.
Krish, welcome to the podcast.
Krish Sathian: Thank you very much. It’s a pleasure to be here.
Steven Cherry: So, what’s going on when we read Elizabeth Barrett Browning’s transcendent metaphor? Are we having a transcendent brain activity or a sensory brain activity or both?
Krish Sathian: Well, I think the answer to that question is really all of the above. So clearly the classical language areas of the brain are active to process what we’re reading or listening to, but the work that you talked about that we just did, as well as work by other people, has increasingly begun to reveal that abstract concepts in general and metaphors in particular are grounded in, as you mentioned, our sensory motor experience, and therefore we also get activity in the sensory parts of the brain or the motor parts of the brain as might be relevant for the concept at hand. And you might also get—I don’t know what the transcendental part of the brain is, I’m not sure people have figured that out—but we might for something that’s highly lyrical or highly poetic or highly emotionally charged, we might get activity in various other parts of the brain as well.
Steven Cherry: So in what way is this a new way of looking at things? How had neuroscientists understood or expected metaphorical language to be processed?
Krish Sathian: Well, so the classical view—and when I say classical it’s really taking a fairly short time frame—the classical view that came into vogue around the middle of the 20th century was that language was pretty much a propositional construction. That is to say, it operated much like a computer language would operate, so you could put various inputs and outputs into this module that controlled language and the processing would come out of that. But actually it’s an old idea that goes back to really the time of Aristotle, that we understand complex things, abstract concepts, by reference to simpler things, concrete concepts. And that would therefore make the prediction that sensory motor areas would be involved. And the pendulum really began to swing towards the end of the 20th century when people started thinking about the brain much more as interaction between different parts of the brain, as opposed to what does this particular part of the brain do. And so our work is part of that, I guess, more recent pendulum swing, in which we are beginning to find evidence for sensory motor involvement in processing things like metaphor.
Steven Cherry: So you did some actual experiments. What did you do?
Krish Sathian: So the experiment that we did was—I should start by saying that we had done some work studying sensory processing per se, and as part of that work we had discovered part of somatosensory cortex, which is a part of the brain that is involved in processing touch, a part of somatosensory cortex is specialized for processing texture in touch. So what we did was, we had people lie in the fMRI scanner while they listened to sentences containing textural metaphors like “He had a rough day” and interleaved those sentences with similar sentences matched for meaning, like “He had a bad day.” And then we contrasted the activity between those two kinds of sentences, and that contrast revealed activity in the texture-selective part of somatosensory cortex, which we had found previously to be the part of the brain known as the parietal operculum.
Steven Cherry: So presumably that part of the brain was not active when the sentence “I had a bad day”?
Krish Sathian: Correct.
Steven Cherry: So, I’m just curious—what was activated in “I had a bad day?”
Krish Sathian: Well, so that’s a good question, but really the way that we use fMRI in this particular experiment and many other kinds of studies—not all, but many kinds of fMRI studies—is really looking at the relative signal, because there are so many parts of the brain that are active during any particular task. For example, the act of, in this case, listening—in other cases it was reading—so the auditory cortex would be active, the general classical language areas would be active, the parts of the brain like in prefrontal cortex, thinking about the meaning of the sentence, would be active, but none of those would be really specific for metaphor or for textural metaphor. So the approach here is in order to try to isolate the parts of the brain that would be specifically active in [relation] to textural metaphors, we would want to have these control sentences which are essentially matched for meaning but don’t have the textural metaphor.
Steven Cherry: Now, I gather it actually took them longer to process the metaphor-laden sentences.
Krish Sathian: That’s correct. The metaphorical sentences took a little longer to process.
Steven Cherry: And they were listening to these sentences. Do you think it would have made any difference if they had read them instead?
Krish Sathian: Probably not, as long as the control sentences were presented the same way. So I think we would probably get the same result, but like with anything in science, I want to be cautious about that because we haven’t actually done that experiment yet.
Steven Cherry: Right. Now, you have a grant from the National Science Foundation to take the research further. What’s next?
Krish Sathian: So, at this point we have tested this idea of domain-specific recruitment of sensory cortex in one particular sensory domain, that of texture. So we would like to extend this to other sensory domains—for example, visual domains—and see if the same kind of principle applies. In other words, how generalizable is our finding?
Steven Cherry: So you might take something like “America is a melting pot” and see if, like, the area that—you know, when people look at an ordinary pot on the stove—was stimulated.
Krish Sathian: Maybe. The idea is that we really want to sort of parse the metaphors that we’re using into different domains because we have predictions for which sensory cortical areas would be activated according to the domain of the metaphor that we test. So for example, regions that process visual shape, regions that process color, like “He was green with envy” or “He went red with rage”—that kind of thing.
Steven Cherry: George Orwell, the novelist and political thinker, wrote a famous essay, “Politics and the English Language,” in which he devotes an entire section to metaphor and particularly what he calls “dying metaphors.” So he says—and this is a quote from the essay—“A newly invented metaphor assists thought by evoking a visual image, while on the other hand a metaphor which is technically dead has in effect reverted to being an ordinary word.” For example, the word iron—and here I’m not quoting—to mean strong or rigid, like an “iron will” or in the movie title last year The Iron Lady. And Orwell says that in between there are what he calls—and I’m quoting again—“worn-out metaphors which have lost all evocative power and are merely used because they save people the trouble of inventing phrases for themselves.” So, for example, we have the expression—this is one that he uses—“toe the line,” in which people sometimes people even write it t-o-w instead of t-o-e, which shows they aren’t even thinking about the metaphor at all anymore. Do you think it would make sense to see what’s going on in the brain here, that maybe there’s a difference between dying metaphors and active ones and dead ones, three different things here?
Krish Sathian: Yeah. So that’s a great question. So, in fact there’s a theory advanced particularly by Dedre Gentner called “career of metaphor theory,” and it’s the idea that metaphors have kind of a life cycle in which, as you pointed out in that set of quotes, that metaphors are initially novel, then they may sort of fade over time. And there may come a time when they’re sort of frozen or dead and nobody really recalls their meaning any more. So that’s something that we actually plan to test in our follow-up studies, to explicitly vary the familiarity of the metaphors. So we would have some metaphors that might be more novel and others that might be much more conventional, and the prediction would be, according to this theory, that the level of activation in sensory cortex for that particular domain would be much greater if the metaphor were novel.
Steven Cherry: So, one more philosophical question if you don’t mind. It seems to me that one consequence of research like this—and as you say, it’s not just yours but other research that people are doing right now—is that there’s less difference than maybe we would have initially thought between having an experience and maybe reading about that experience in a novel. And you know, the 18th-century philosopher David Hume held that the only difference between experience and imagination—he spoke of “impressions” versus “ideas”—was really their intensity or vivacity, and it seems like a lot of fMRI research nowadays seems to be consistent with that. Are neuroscientists becoming Humean idealists?
Krish Sathian: Well, I think that’s a great question. In fact, you know, not only neuroimaging research but also psychological research is certainly heading that way. This idea has come to be known as “grounded cognition,” and people sometimes refer to the scientists who are espousing this view as “new empiricists.” And the idea is that in order to understand other people’s actions or experiences, in order to remember what you experienced yourself, you actually run simulations of those experiences in your own brain, and those simulations either help you to recall that experience in the case of memory or understand other people’s actions or thoughts in those particular contexts.
Steven Cherry: Well, Krish, in my highly biased view, I can’t think of a better use of our tax dollars than to see if science can understand what’s going on when we use language and maybe to see if we can’t use it better. Although if everyone could use metaphors perfectly, I’d probably be out of a job, by the way. But anyway, good luck to you, and thanks for joining us today.
Krish Sathian: Thank you very much. It’s been a real pleasure talking to you.
Steven Cherry: We’ve been speaking with professor of neurology Dr. Krish Sathian about some new discoveries in the way the brain processes metaphorical language. For IEEE Spectrum’s “Techwise Conversations,” I’m Steven Cherry.
Announcer: “Techwise Conversations” is sponsored by National Instruments.
NOTE: Transcripts are created for the convenience of our readers and listeners and may not perfectly match their associated interviews and narratives. The authoritative record of IEEE Spectrum’s audio programming is the audio version.