Functional Specificity: What it means and what it doesn't (17:53)
April 20, 2016
July 29, 2015
All Captioned Videos CBMM Summer Lecture Series
Nancy Kanwisher, the Walter A. Rosenblith Professor of Cognitive Neuroscience in the Brain and Cognitive Sciences Department at MIT, talks about the history of attempts to find functionally specific regions in the human brain, what it means for a brain region to exhibit functional specificity, the important role of fMRI in the study of brain function and use of functional Regions of Interest (fROIs) in the analysis and interpretation of fMRI data, and what aspects of the human brain enable uniquely human abilities such as language understanding.
NANCY KANWISHER: So I've been talking, just very crudely, about the idea of brain functions living in particular-- mental functions living in particular parts of the brain. What exactly do we mean by that? Let me just first state it more specifically, OK. So here's me and my special bits.
Anybody has pretty much the same special bits. There are icons on the left, show you what each of these things do. And the basic claim is that the human mind and brain contains a set of highly specialized components, each solving a different specific problem. OK. That's the idea of functional specificity in the brain, OK.
And that's interesting because each of these regions is present in, more or less, the same place, in every normal person. It's part of the basic architecture of the human mind and brain. This is, in some deep sense, who we are. OK, that's the basic idea. But that idea is often confused with a whole bunch of other ideas. And I want to go through them and talk about how they're different.
OK, so we can think of this as different questions we can ask. The functional specificity question is just, as I just said, the question of whether a particular brain region is only or primarily engaged in a specific mental function. OK, call it X. That's what we're talking about.
That's different from, what you might call, anatomical specificity. The question you ask about anatomical specificity is, is this the only brain region that carries out that mental function? That's a different question, right. So to see that, let's look at a diagram, right here, of the bottom of the brain and the side of the brain, with just a few of those functionally specific regions.
And again, this is highly schematic. Nothing's a neat little oval. Reality is messier than that. But this is an idealization. And there are at least three face-specific regions in the brain that have been identified, probably lots more.
But this one is distinct from that one. They're both quite face-specific. And my point is very simple. The fact that there are two of them, doesn't in any way undercut the functional specificity of either one of them.
OK, so there's two different questions. Is there just one region? And for any particular region, is it function specific? Often people say, oh, you know, there's no specificity for faces. There's a bunch of regions. It's like, hello, you know. Anyway-- OK.
Oh, and a particular terminological problem that arises, is people use the word distributed in two very different ways. The original use of the word distributed, as in face representations are distributed over the brain, OK. The original meaning of that is an argument against what I'm saying here. That is, essentially, the idea that face representations are spread so broadly that they're on top of representations of objects and words and places and everything else, OK.
So that's an important counter argument to the idea I'm putting forth here. But then people sometimes say, well, there's three face areas. They're distributed. That's a confusion, right, that they may be three areas, and they be spatially in different parts of the brain, but they're not distributed in the sense of on top of other things, OK. All right.
Next question, necessity. OK, so with functional MRI, you can ask what regions turn on when you do different mental functions, OK. With functional MRI, you cannot answer the deeper, more important question of which regions are necessary for different mental functions, OK. This is really of the essence.
I mean, actually, if we had methods that worked as well as functional MRI, that could tell us about necessity rather than just what turns on, we would all be doing that. Because that's the deeper question. That's the causal question. What regions are on the causal chain in face recognition, not just what regions happen to turn on when you do face recognition.
We want to know, if you took that bit out, would you have a problem face recognizing faces, OK. So this question is tightly connected to the question of functional specificity. And in fact, I would argue that it's something we really want to know, right. Often we can't tell because we don't have the right methods for it, but it's what we really want to know, OK.
There are other methods that can get at this, so I won't go into them now. But if you watch on my website, on nancysbraintalks, you can see me getting zapped with transcranial magnetic stimulation. That's a causal method that you can use to test causal roles in parts of the brain. And there's also a talk on there on what happens when you stimulate the face area, using another causal method of electrical brain stimulation, OK.
Next idea, sufficiency. OK, so there's necessity and sufficiency. The question here is, is that brain region, in question, is it sufficient for the mental process, OK. Well I would say this is just a wrong question to be asking because nothing is ever sufficient, right. Right? It's actually-- philosophers write about what a confused idea it is.
I won't go there but the problem is, OK, with sufficiency, basic idea is, OK, let's take this diagram. Let's choose one region, like that. OK, that region responds selectively when you perceive the spatial layout of places, OK.
Right now, as I look around in this room, this room is longer on this axis than that axis. That's being represented, chronically, all the time as I lecture, in that part of my brain, OK. And yours too. Whether you're thinking about it or not. Actually, we haven't tested that. I'm pretty sure it's representing it, whether you're thinking about it or not. But it's an interesting question.
OK, I lost my train of thought. Right. We can then ask, is that region sufficient for representing spatial layout? But that's ludicrous. How could it possibly be, right? No region is ever sufficient for a mental process. At the very least, you need to get the information into there. That's a visual region.
You need to get-- you need eyeballs. And you need pathways between your eyeballs to get it up there. And further, if you had all of that-- suppose you had all the inputs right, and you had eyeballs looking out at the world and that region is working fine there.
And then suppose you surgically excised that region with its eyeballs and kept it alive in a dish, and you aim the eyeballs around with the world. And you asked what was going on in those neurons. Would that be a representation of space? I mean, would that piece of tissue be aware of space?
| philosophers write about that. I don't think so, right. To be aware of something, you need to connect up to the rest of your brain, right. Other parts of your brain need to have access to it.
That little region isn't sufficient for the whole perceptual experience, presumably, right. OK. All right. This is just to say, sufficiency is just a question that shouldn't come up. It's a confused notion that doesn't get us anywhere. But it does come up. All right.
Next, interaction and connectivity. OK, the question here is, is this particular part of the brain that we're interested in, part of a network? Or in another way, does it interact with other regions?
OK, so we go back to our diagram. Here's a bunch of regions drawn here as little isolated universes. But of course they're not isolated universes. As I just said, none of them could do anything without inputs and outputs. So of course each region is part of a network. Like, duh, OK.
Who knows what the connections actually are. Our methods for discovering them are kind of lousy. Here's some made up ones. Of course each region interacts with other regions, right. They have to, otherwise we couldn't use the information and act on it, right, or think about it.
My point is just that, in no way does the fact that a given region is part of a broader network, undercut the possible functional specificity of that region, OK. This is something that's widely-- people are confused about, colleagues of mine who should know better. One wrote a paper called "Let's Face It, It's a Network", talking about face areas, as if that was somehow a challenge to anything I or anyone else has said about the specificity of face regions. Of course it's not a. Challenge of course it's a network, right. OK.
Sorry. You guys are inheriting my annoyance with my colleagues. It's not your fault.
OK, the final one, which is probably the most common confusion, is innateness. OK, so the question here is, how did that candidate region of the brain, that we're thinking about-- how did it arise developmentally? Is it innately specified? Is there a complete description of how to make that whole brain region in the genome, or is an important part of its selectivity derived from experience.
OK, so this is a classic question for people who have asked themselves, for hundreds-- I'm sure, thousands of years, OK. I mean, it's a fundamental question. People ask themselves that question because it's fundamental. And we all desperately want to know the answer.
Unfortunately, we don't have good methods for answering it, right. We try. We do like weird stuff around the edges, but basically, we can't answer this question yet. But it's widely confused. And I'm sure you've seen cases of this. It comes up all the time.
One of the times that was most salient to me was in 1991, the British neuroscientist, Simon LeVay, published this finding that just kind of electrified people. He showed that part of the hypothalamus is different in gay men from straight men, OK. And immediately everybody said, that shows that sexuality is innately determined.
Well, no, it does not show that. It shows that there is some biological correlate in the brain. But you can find all kinds of stuff in the brain that's a result of experience that may not come from genes directly. OK, so this is a common confusion between having a biological correlate in the brain, and being innately specified in the genes and inevitable from your genome, OK.
So just to underline that point, I'm going to give an example of a very functionally specific brain region that absolutely, definitively was wired up by experience, not by genes, OK. And that's what's often called the visual word form area. So here's the background. This is a paper that we published quite a while ago, with my then postdocs, Jia Lia and Chris Baker. And so we wanted to know whether these specialized regions of the brain can ever gain their specialization from experience alone.
It's very hard to tell for faces, places, or bodies because any human subject you scan has been looking at faces, places, and bodies their whole life. And so have their ancestors. And so how the hell are you going to tell whether the thing got wired up from the experience of their ancestors via natural selection and DNA and so forth, or whether it got wired up in that individual's experience. God help you.
People are doing these things in monkeys and other animals, and they're making bits of progress, but it's difficult. And it's pretty much hopeless in humans. But in this one case, we can answer the question. His question is, can experience alone, without a specific genetic blueprint, ever create a specialized region of the brain, OK.
OK, so the key test case is visually presented words. OK, so why is this a key test case? This is an argument that many people have made before me, most notably Martha Farah at UPenn, who made this case very beautifully, and is what motivated us to do this study. She pointed out that our experience with visually presented words is on a par with our experience with faces, right.
Maybe not exactly the same, but we have loads of experience and it goes way back. And we look at words all the time, right. Yet, crucially, people have only been reading for a few thousand years. And that means that's not enough time for natural selection to have crafted a specialized circuit just for reading.
So that means that if we find one in the brain, it was probably wired up by that individual's experience with reading, not by their ancestors via natural selection and evolution and genes. OK, everybody got the argument? OK, so that's why this is such an interesting brain region to look at. And so we set out to look for it. Many other people have looked for it.
But as a sidebar, there's a pervasive sloppiness in the field where most of the other work compared-- look for brain regions that respond more when you look at a word, than say, when you look at a flashing checkerboard. And they found some activations. They said see, it's a visual word specific region.
My reaction is well, maybe, maybe not. And let's do it right. OK, so how do you do it right? You compare the brain's response when you look at visually presented words, not just a flashing checkerboard, but lots of different things so that you can really test, in a serious way, how specific it's response is.
OK, so here's what we did. We started by comparing the brain's response when you look at visually presented words to line drawings-- black and white line drawings of familiar objects, OK. And actually, the words named the same object, so they were the same concepts. They're both complex, high-level, visual stimuli, but some are words and some are pictures of objects. OK, so that's a crude contrast, but it's better than words and checkerboards.
And when you do that, you find, in most subjects, a tiny, little region, it's about a tenth the volume of the fusiform face area, that responds more to words than line drawings. OK, so that's promising. But to test it more seriously, we tested a bunch of other conditions.
OK, so here's what we did. We scanned subjects who are readers of English, but not of Hebrew or Chinese, and we show them a bunch of different kinds of stimuli. OK, so English words consonant strings, OK. Chinese words that are unfamiliar to them. They don't speak Chinese. Hebrew words that are unfamiliar to them, digit strings and line drawings.
OK, so we look in that candidate word selective region and we say, is it really words selective, OK. And here's what we find. OK, so here's the magnitude of response in that region, to these conditions. And what you see is a high response, both to consonant strings and to words, OK.
So that immediately tells you, it's not words, per se. That brain region doesn't know the difference between a real word, like table, and a non-real word like "mafer", OK. It doesn't even know the difference between table and [? Fubul, ?] whatever that is. OK, so it doesn't even know about which letters can legally go next to each other. Actually that's debated. I think we're right, but there's some room for debate on that particular question.
OK, in any case, it responds more to things with letters, than to anything else. And importantly, look at these visually, pretty similar things. The Hebrew word is pretty visually similar and the response is much lower. Digit strings are damn similar to words and letters and the response is much lower. OK, and line drawings are way down there, as are Chinese characters, OK.
So that's pretty suggestive that that region gained its selectivity from these subjects' experience. But it doesn't 100% nail it. This is like a A minus argument, I would say. To really take it all the way, we need to vary experience.
OK, so we can't find people in Cambridge, who are otherwise normal but who can't read English. If they can't read English, there's something else going on. Well, I guess it could be non-native speakers who just arrived here. But easier, is to find people who read English and Hebrew, but not Chinese, OK. Loads of those around, we can scan them. And we did.
And here's what they look like. Here are the English words and consonant strings. Those are the Hebrew words. And here is everything else, OK. So only difference between these two groups is these people have long term experience with Hebrew and those people don't. OK, so that nails it.
This brain region reflects that individual's experience and is crafted for a very particular, selective response to stimuli experienced by that subject. OK, everybody get that? OK, I'll get to your question in just a second.
OK, so this is a kind of functional specificity in the brain that cannot be innately specified, right. And so I would say, here's a case that has to be due to that individual's experience. That is true for that one region. It doesn't mean that the face, place, and body areas necessarily result from that individual's experience. Those ones could be all, or in part, specified in the genome.
It's just an existence proof that, for at least this one region, where we can test the question, that region was wired up by experience, OK. So that's my list of the distinct concepts that are often confused with functional specificity. Innateness is a totally different question and orthogonal from the question of functional specificity.
A functionally specific region could be innate or not. It's a different question. All functionally specific regions are surely connected up as part of a broader network, duh, and so forth.
Associated Research Thrust: