Speaking of Psychology - Why can some people speak dozens of languages? with Ev Fedorenko, PhD

Episode Date: June 22, 2022

For those of us who speak only one language, the idea of learning twenty or thirty sounds impossible. But there are “hyperpolyglots” who have managed this remarkable feat. Evelina Fedorenko, PhD, of the Massachusetts Institute of Technology, discusses what sets polyglots apart and what scientists might learn from studying them. She also discusses how language is processed in the brain, why it’s so much easier for kids to learn languages than adults, the relationship between language and thought and how we can think without language, and more. Links Ev Fedorenko, PhD Speaking of Psychology Home Page Sponsor Newport Health Learn more about your ad choices. Visit megaphone.fm/adchoices

Sponsored by Newport Healthcare, providing results-driven mental health care for teens and young adults ages 12 to 28, who are struggling with trauma, depression, anxiety, and other mood personality and co-occurring disorders. At Newport Healthcare, tailored treatment plans foster sustainable healing to help young people move from struggling to thriving. Learn more at newporthealthcare.com. For those of us who speak only one language, or maybe even speak two or three, the idea of learning 20 or 30 sounds impossible. But there are people who have managed this remarkable feat. They're called hyper polyglots. There's no official definition of hyper polyglotism.

The International Association of Hyper Polyglots welcomes members who speak six or more languages, but some hyper polyglots speak dozens with varying degrees of fluency. So what sets hyper polyglots apart? Are they simply smarter than the rest of us, or harder working, or more motivated to learn? If you put them in a functional magnetic resonance imaging scanner, do their brains look different from those of monolingles or bilinguals? And what might scientists discover about how the brain processes language by studying people with remarkable language abilities? Welcome to Speaking of Psychology, the flagship podcast of the American Psychological Association that examines the links between psychological science and everyday life. I'm Kim Mills. Our guest today is Dr. Ev

Federenko. Dr. Federenko is an associate professor in the Department of Brain and Cognitive Sciences at the Massachusetts Institute of Technology, where she explores how people understand and produce language. She uses brain imaging and other research methods to study healthy individuals and people with language-related brain disorders, as well as hyper polyglots. Among the questions she's interested in are what's the relationship between language and other cognitive abilities? Can we think without language? What is the organization of the language system in our brain? And how does it emerge in childhood and change over the course of a lifetime?

We'll try to touch on those topics today. Thank you for joining us, Dr. Federenko. It's my pleasure to be here. So, as I indicated, your research is much broader than hyper polyglotism. But let's start with that topic because it's so fascinating. How did you become interested in studying people who speak? dozens of languages and why does studying this extremely rare ability, why is this a useful way to learn more about language and the brain? As you know, for many years, centuries really, researchers have studied individuals with language problems.

And the intuition there is very clear. We can learn a lot about how something works by looking at ways in which it can break hormonal function. And we have learned a lot from individuals with both developmental disorders like autism as well as acquired disorders like aphasia due to stroke or degeneration or head trauma or something like that. But there is this other extreme in any cognitive ability or perceptual ability you can think of. There's this extreme where it seems that some tail of the population, some small number of individuals in the population are especially good at something. And it's kind of interesting to think about what it means to be good at language, and there's probably many different ways in which you can be good at language. But it certainly seems that this ability to acquire multiple languages, multiple mappings between meanings and forms like words and constructions is certainly one way in which you can be good at language. And some people seem to have a propensity for this, or at least in one way or another, become hyperpholyglots. And so we get interested and maybe we can learn something about how the typical system works,

but also studying this other extreme of linguistic abilities. Do we have any idea how rare hyper polyglotism is? And are there particular characteristics of hyper polyglot, such as, are they mostly men? Do they tend to be ambidextrous? I mean, what's unique about them in addition to their language ability? The prevalence is very hard to estimate. There are now a few of these organizations that try to kind of bring together polyglots, but I still think there's many people who kind of do this on their own. And

Vaughn Smith, who recently was covered in the Washington Post article, is one of those people, right? That he just does it and nobody knows he doesn't belong to any societies or anything like that. So I think it's probably whatever estimates exist out there are probably underestimates. In terms of characteristics, it's a surprisingly heterogeneous population. One thing that seems to hold for many people is that, and this was counterintuitive to me, is that they're not typically individuals who grew up in multilingual societies. So there are some countries like, you know, the Netherlands where most kids grow up speaking three languages or Switzerland or, you know, parts of India. They typically grow up in relatively monolingual societies or societies where there's

one dominant language, like Russia has a high proportion of polyglots. The U.S. does as well. So they grow up monolingual and then at some point they just get interested in learning languages and maybe start with like a typical class in school and then realize that they like doing this or are good at it and then just kind of keep going with this. But otherwise it's a very heterogeneous group of people. Some are really extroverted and can learn through kind of social interaction with native speakers. Some are quite introverted and learn primarily from books. It seems that the ways in which they're good at learning languages is also different. We can talk more about that. But yeah, I think more of the people that we've tested so far are males,

but I would not attribute much to that just because historically in our society, males have had more encouragement for certain things. And I think there's still biases these days so that girls may be not as much encouraged or something like this. And so I would be very cautious in making any claims about gender biases until our society is healthy enough in terms of the lack of sexism when we can assess these things more fairly. Yes, we're waiting for that day. So much of your research uses fMRI brain imaging. What have you found when you look at the brains of hyper polyglots? How do their brains look or function differently compared with the brains of people who only speak one or two languages?

So, of course, for this comparison, if we're comparing polyglots and typical individuals, we have to look at the responses at these participants' native language, right? Because otherwise, it's difficult to make comparisons because there will be differences in proficiency. But I can take you, your native speaker of English, and I can take a polyglot, whose native language is also. language, right? And if you look at the neural responses to native language, we find that in hyperpolyglots, the language system looks smaller than the typical individuals, meaning the amount of territory that kind of lights up, quote unquote, in the brain when they process language is smaller, and the regions that respond to language respond less strongly when they process language

compared to you or me. And this was originally an accidental discovery. We had somebody contact us a bunch of years ago now and said, I have a son who is a polyblood, learning, you know, I had like 20 or something languages. Would you be interested in testing him? And, you know, I was young. I guess I'm still pretty young, but I was younger

and very interested broadly in how language works in the brain. And so I said, yeah, sure. we had some extra resources at the time, and so we brought him out, and we scanned him on our usual paradigms. And lo and behold, his language regions look really wimpy. And I was like, that's strange. You might expect a priority, right, that somebody who is really good at language will have these robust, strongly responding areas. And the responses look small. And we tested them again the next stage to make sure things are replicable. And again, they look small. And I'm like, oh, that's, that's interesting. And then I started talking to other people working in different

domains. And people brought up examples of motor learning. And they said, well, actually, that's kind of what happens when you learn new motor skills. So if I teach you some complicated motor skill, like some tracing of complex patterns or something, and I have you do it initially, you'll have big chunks of your motor cortex respond as you're still kind of getting better. And as you get more and more proficient, and it becomes kind of more automated, the amount of motor cortex that you have to recruit becomes smaller. So you don't need to use as much tissue to achieve the same level of performance. And so maybe something similar is happening here.

So that's the parallel we drew in interpreting these results. Oh, yeah, and I should say that, you know, then eventually we kind of got other polyglots contact us. We found a few local ones at MIT in surrounding area. And so in the end we ended up with like a set of 20. that finding holds across them. So looking at the distribution of the sizes of the language areas in that population is definitely shifted to the smaller side compared to this large normative scent that we have now like hundreds of typical participants. So what else sets hyper polyglots apart? Are they generally extremely smart in other ways? Or is

their talent just restricted to language? Do they have to study hard to learn languages in the that other adults do or does it come more easily and naturally to them the way that it might to children? In the study that in the one study we published on the neural responses to native language in polyglots and typical individuals, we actually matched for IQ. So the difference that we observed holds in spite of similar levels of kind of general fluid intelligence. We also looked at the responses, neural responses in the brain system that kind of supports

general fluid reasoning and those regions look similar in size. So it's not like their whole brain is efficient, no matter what system you look at, it seems specific to language. And in general, we and others have found that those two systems are quite separate, the system that supports thinking and reasoning and the system that supports language processing. But in terms of what makes them different, I don't think we know quite yet. So it's undeniable that there are some people for whom learning languages is easier than others.

That just that variability is there. I myself learned a lot of languages until I moved to the U.S. And it all got kind of dormant and forgotten. But growing up, I learned it. And I loved doing it and it was easy. Like, I enjoyed doing it and it kind of came naturally. I could remember things easily. I could decipher grammatical patterns.

So, and you'll hear other people talk about it. And they're like, yeah, it seems like I get like a good feel for a language very quickly. Like I can kind of understand which words I relate and construct things once I learn a few words, construct things much more easily than my peers seem to they're learning me in the same environment or something like that. But there's also, again, quite surprising that there's also hyperpolybloods who are pretty adamant that they don't have any special talent. Again, maybe introspection is not the best way to, but they just say, I just like doing it. It's like a, it's like a hobby for me. I spend a lot of time doing this.

and I get better, but I don't think I'm linguistically gifted in any way. I just kind of spend a lot of time. Like, I spend, you know, 20 minutes each day or whatever, you know, an hour a week on each of my 20 or 30 languages. And I get good because people get good with practice, right? Like, anybody will get good. Almost anybody will get good with practice on almost any task. But, of course, there is this underlying variability. And it's possible that some of the kind of more extreme hyperpolyglots who achieve proficiency

in upward of 20 languages, maybe it's some combination of some innate talent, if you wish. People sometimes get upset about the use of terms like that, but, you know, whatever, innate biases to be good at language, whatever, and a lot of practice. And then maybe combine those things together. You can kind of get to the real extreme of that ability to learn so many different mappings between meanings and forms.

But I think there's a lot we don't have. understand this is a very understudied population. There's not many of them, and they're not like all in one area or something. So it's a pretty challenging group to study. But I'm hoping that some of the stuff we've been doing recently will inspire others to try to characterize them kind of more broadly and in all sorts of other ways, aside from just your responses. So I thought it was interesting in preparing to talk to you that there isn't a connection

between, say, musical ability and language ability? Because you would think there are some languages that are very tonal and they almost sound like music. And yet there isn't a connection. Do you have any understanding of why that might be? It just seems counterintuitive. To clarify the lack of, so the dissociation between language and music is happening at the level of kind of language understanding.

So the system that you use to understand and produce language, so to kind of infer meanings from sequences of words and convert your thoughts into sequences of words, that system is totally separate from music. And this is contra many proposals. And that is now, I think, we and a few other groups have strong evidence that the language regions are just not working at all when you're listening to music, even if it's complex and structured and has all sorts of properties that people argue maybe shared between music and language. But there is some overlap at the lower level auditory processing.

So there is, for example, a part of our auditory cortex that cares about pitched sounds. So any kind of pitch sounds, it can be a train horn or a speech segment or a piece of music. And anytime there is kind of pitch modulation, harmonics present, that region is active. And so through that shared machinery, there are indeed some transfer effects so that musicians, for example, have an easier time learning tonal languages. those effects have been reported for quite a while. It doesn't seem that being a musician, so having a lot of practice with music or being musically talented or having things like absolute pitch makes you better at learning languages in general, but it will help with the aspects of language that rely on that ability that

you're training up, which is paying attention to pitch in the signal, but it doesn't see like a general thing. Why is it, do scientists know why babies and children can learn language so easily and that it becomes more difficult as we get older? I'm sure it's hard for you to get little babies and put them in fMRI machines, but I know you've looked at some kids. Yeah, that's a million dollar question that is still very much debated. So a lot of things change when you're learning a language as a kid and you're learning a language as an adult. So for one, when you come into the world, you don't know anything about the world, right? So at the same time, as you get it's starting to get exposure to speech, you're also learning

all sorts of things about the world, the social agents in the world, the physical properties of the world, the objects, how they interact. And at the same time, you're hearing linguistic signals or watching them if you're a baby learning sign language. And then at some point, you're, you know, as a baby, you start noticing that there's some non-randomness so that like every time you know your daddy says ball there seems to be this round thing around right well you don't know the word for round but you kind of know just that there's a thing like that looks like this round thing around and eventually i think it must click for babies

that there is this non-trivial correspondence between these things that we're making the sounds that we're making with our mouths or with our hands if we're a signer um and the environment And then once that clicks, then I think, you know, like language just kind of goes through the roof. And word learning happens and eventually they start composing words and so on and so forth. Why we lose the ability to learn language as easily, like I said, very debated. So for one, you already have a set of mappings, right? You've learned a language. So there's already a fixed system there of how things map.

So learning a system on top of that, another set of mapping on top of that may be harder. I always make this point that I haven't heard others focus on as much. But the way that you learn a second language is typically very, very different. So we actually don't have very well-controlled experiments of similar environments for first and later acquired languages. Because, you know, if you've had kids, I have a four-year-old, so we just went through this whole language learning thing. if you're a baby, every time you're awake, there's a face in front of your face and it's making speech sounds at you. Right? Like, and it loves you.

Like all these caregivers around you, they love you and they interact with you every time you're awake. Like pretty much every time you're awake, there is some kind of language. It's a very different experience from taking a class. And even in immersion, right? Like you may go to Paris. You're sitting in a cafe. Somebody will occasionally talk to you. But it's not like you're constantly having people who love you.

try to teach you the skill so that we can interact better, right? It's just a very different, it's a very bad, poorly controlled comparison. And there are some claims that underdeveloped executive abilities early on may be helpful, that maybe interfere later on. There is some claims that babies initially don't have the propensity. So speech is a continuous if we're talking about spoken language, right? there's no boundaries between words. And so kids will kind of hear continuous chunks of speech, sometimes in pretty large chunks.

They don't know how the word boundaries are within that chunk. And people like in Bal Arnon in Israel has argued that actually not always parsing things into individual words may help you initially bootstrap the system because you're learning bigger chunks from which you can potentially more easily infer. kind of multi-word relationships and things like that. Because once you're learning, once you're grown up, you understand words, right? You understand that there's words. And so when you're learning a second language,

you learn everything in individual words as opposed to paying attention to like phrases and constructions where you may benefit from the fact that a certain sequence of words always happens together. So there's all sorts of hypotheses out there. In terms of the evidence, kind of the largest scale studies that do exist, do suggest that there is a drop in your ability to learn languages in a native-like way

after around puberty like 15, 16 kind of late puberty. But again, there's only some abilities that are affected. So grammar and sensory motor stuff, so accents, accent is something that's really hard to lose after that time period. In terms of learning words, you can kind of keep learning words no matter how old you are. Right. Yeah, so that's where we are. So still a lot of work to be done. Sponsored by Newport Healthcare, dedicated to providing sustainable healing for young people struggling with mental health concerns so they can move from self-destruction to self-esteem. Based in compassion and driven by outcomes,

Newport serves teens and young adults ages 12 to 28 at residential and outpatient programs nationwide. through tailored treatment and family support, Newport successfully addresses psychological and behavioral health issues, including depression, anxiety, PTSD, personality disorders, and substance abuse. Learn more at Newporthealthcare.com. So going in a slightly different direction, you also study people with brain abnormalities that people might think would hurt their language abilities. For instance, you recently published a paper about a woman who is missing her left temporal lobe, and yet she's remarkably unaffected by it. She has totally normal language abilities, even exceptional in some ways because she speaks a second language fluently. So what do case studies

like that tell us about how language is processed in the brain? Yes. So this is also another line of work that kind of came about somewhat accidentally because this individual reached out to us and volunteered her brain. So she didn't know that anything was different about her brain until she was in her 20s and she kind of accidentally found out. And it turns out that it's actually not so uncommon as you might think to have an unusual brain. Because our brains have a lot of redundancy and plasticity, especially early on, a lot of early damage goes undetected. So, some big event may happen in your brain as you're going through the birth canal or shortly thereafter.

And you have so much brain tissue to work with that you grow up totally normal. Sometimes people kind of live their life happily and don't know that they have some big chunk of their brain missing because they don't have any symptoms. And that means they don't go to a doctor and they don't get referred to a neurologist or anything like that. There is also evidence from surgical interventions that sometimes are done early on for cases of, for example, severe epilepsy, where I don't think they do that quite anymore as severe, but it used to be the case that sometimes whole hemispheres would be removed.

So hemispherectomy is a surgery that used to be done. And again, those kids, if the surgery is done early enough, they grow up and they're relatively fine. Sometimes they'll have some motor deficits. But in terms of their cognition, everything seems okay. Now, of course, this is in striking contrast to having a stress. stroke later in life or having a surgical procedure that removes some part of your language cortex later in life, which will lead to problems.

And I think you can learn a lot of different things from this population. So this paper we published, we asked the question that you couldn't even really ask in any other way. We asked about whether, so the language system consists of temporal and frontal areas. The frontal cortic layers, like in the front of your head and side of your head, lateral temporal side of the temporal lobe. And typically we see them both as early as can be measured. As you said, it's hard to put kids in the scanner. So you can do it from about four or five pretty well. At that age, if you look at their language regions, they have the temporal regions

and they have the frontal regions. The question is, what is the relationship between them? Do some of them emerge earlier than do the frontal regions emerge later because there is also a lot of evidence that frontal lobes in general are slower developing. And so this individual who grew up without her temporal lobe, left temporal lobe, could allow us to ask, would there be responses, neural responses to language in her left frontal cortex, which is perfectly healthy? A lot of other stuff is happening there. So would there be some language responses in that part of her brain?

Then there isn't. So her language system was totally remapped onto her right hemisphere. I shouldn't say remapped because there was also a claim that language is originally bilateral, early on bilateral, and then it becomes more lateralized as you grow up, although that's also controversial. But in any case, her language system is happily living in her right hemisphere. But unlike in typical individuals where, say, if I put you in the scanner, you likely will have your language areas in the left. But you'll also have some responses in the right homotopic corresponding areas. That's kind of what you typically see. And in EG, no matter how much this is this individual without her left temporal lobe, no matter how powerful the tests were that we were using to detect language responses, there's just zero response to language anywhere in her perfectly healthy left frontal lobe, suggesting that perhaps the lack of this temporal cortex is what's making these regions not emerge, right?

So maybe you need these temporal to frontal connections early on to kind of wire up the frontal parts of the language system. But ever since this word got covered by the media, I've had a lot of people write to me and say, you know, I also have an unusual brain and that will send me like screenshots of their MRIs and they have like their right temporal lobe missing

or their frontal cortex chunks missing or their friatal lobe missing. And so one question that I think we can very interestingly address in this population is the relationship between language and social cognition. So language, as we mentioned, is typically in the left hemisphere, present more strongly in the left hemisphere, and all sorts of social functions are more strongly present in the right hemisphere. Now, if you only have one side of your brain, or largely, you know, if a lot of the brain is missing on one side, some parts of the, say, social cognition would have to move to the left

or some parts of language would have to move to the right. And we want to understand the constraints on how these two systems can coexist within the same hemisphere. Like, can they start sharing resources? Can they actually recruit overlapping bits of the brain and still manifest as typical cognition with no deficits in either language or social processing? Or do they just take up less space and remain fully segregated? Or how do they reorganize in such cases? So that's where we're going with that work. So when you looked at EG, the woman with the missing temporal left temporal lobe, did she have social deficits?

Were you looking for that? She doesn't. I mean, we've tested her on some basic stuff, and we even tested her in a pretty extensive battery of so-called pragmatic tasks. So these are abilities that are kind of at the junction of language and social cognition, so understanding sarcasm or understanding when somebody is asking your question, but really they want you to do something. right? If I say, can you please best the salt? I'm not really asking you if you can do it. I'm asking you to do it. That's just the way we do this. Yes. And she looks absolutely typical on this, you know, or like also things like using contrast of prosody, right? Like if I'm

emphasizing a sort of part of my message to indicate that something is different from what you might have thought, like, no, I want a red dress, right? Suggesting that maybe we were talking about a blue dress before and I'm contrasting this. So these kinds of cues that are at the junction of understanding people's mental states, like intense, like, why are they saying it? Like, why are they saying it in this way? And linguistic processing seemed totally intact at her. I think we can maybe look at more stuff, but really she's like a delightful person.

And, you know, we've now interacted with her for, she's come out, I think, three times. and you know, you would never suspect that anything is different about her brain. So it also tells you that we don't need as much brain as we have. Right. And of course, the surgeons would be out of jobs if that wasn't the case, right? They'll tell you that there's a lot of stuff you can cut out, and even in adults, and we're totally fine. One question you look at that I find really fascinating also is the relationship between language and thought

and for many people, it seems intuitive that you couldn't engage in complex thinking if you didn't have language, but you've found that's not necessarily the case. Can you explain how you can have thought without language? I think intuitions about our own thinking processes have their limitations. We may have some intuitions about how we think, but really I think empirical data are more compelling. And so this question of whether you need language to think complex thoughts is a really big and deep one. And there is a very strong propensity historically to link them together. Because basically the way it goes as, oh, you know, here is a species of animals.

Right. Humans. They're really smart and they have language. Therefore, they're smart because they have language. And that logically just doesn't follow at all. it could be that they got smart, they got smarter than other animals, and then the communication system kind of caught up. Once you have enough mental complexity, enough complexity in your thoughts that you want to share them,

maybe you then come up with a more sophisticated code so that you can share your thoughts with one another. But in terms of evidence, yes, it seems that the language system is relatively silent as measured with tools like FMRI, when you engage in playing chess, solving problems, doing math, all sorts of things that you would think would be closely linked to language or relying on linguistic representations and in individuals with severe aphasia,

so severe language problems, you can ask them to do all sorts of things and short of translating them into language, like understanding language or producing language, they can do everything that you and I can do. They can navigate the world. They can understand others' intentions, as long as you don't have to rely on the verbal cues, right? You can infer it from people's behaviors, actions, facial expressions and things like that. They can do Sudoku puzzles. I mean, they can basically, they have the same richness of their mental worlds as they did before they had this major stroke event. They just lose the ability to

convert those thoughts into a code that other humans can understand. But they have language, right? I mean, they could speak before the stroke, so they actually have language, as opposed to, say, like the Helen Keller test, right, where she was, you know, a little girl before she actually got language as we understand it. That's right. So the question is they're like, was it necessary for language to be there early on to get some of these capacities to develop? And that's an interesting question because, of course, in these adult stroke cases, you know, everything was they had both language and they had their mentally, you know, general intelligence. And then at some point, the language system was basically taken out and the rest of cognition seems okay. But maybe if you don't have access to language early on, you can't actually develop the ability to, you know, categorize things in the world or do math or something like that. And of course, it's much harder to ask that question because you can't deprive kids.

of language, but there are cases where deaf kids are born to hearing parents, which is actually the majority of deaf kids, are born to hearing parents. And so their parents don't know sign language because they typically wouldn't have had a reason to learn sign language. And in some such cases, these kids don't get access to language for sometimes a few years, sometimes into their teenage years. It hardly ever happens in the US anymore because of all the education about deafness and the importance of sign language.

From the anecdotal evidence and some sparse experimental evidence that exists, it seems that again most things you can learn without language, not that it, you know, not that it wouldn't be easier if you had language, but it's not like you can't learn math absent linguistic input. And I think there's more work to be done on that. I think it's a very exciting population that can shed light on this question, but there's only a handful of individuals who really were lacking language for a long enough time that it would have mattered. But I'm trying to team up with a researcher at BU to try to look at this more systematically. Another area where you're doing research is on artificial intelligence language learning systems.

Do AI language systems learn language the same way that humans do? Almost certainly not. I mean, they're basically, they have a bunch of text. jumped on them, right? That's not at all like we were saying, you know, the way a baby learns is very, very different. I mean, and there's many ways in which it's different. Like, one, you just get linguistic input and it's completely unrelated from the perceptual and motor experiences in the world, right? They just have sequences of words as all they're being fed, the models. And of course, they also don't have any social interactive component, right? There's no way they can understand intent. behind why somebody might say something. All they get are co-corrences among words. Now, with a large enough training set, you can get a lot from those co-currences. Language is very powerful in that it reflects all sorts of regularities about the world. And so in some cases, if you have a model generates some coherent text, it's very tempting to think that the model knows something, quote, unquote.

But of course, it doesn't really, all it knows is linguistic regularities. and it so happens that they reflect some world knowledge, but it's all restricted to the knowledge of language. The fact that they learn differently doesn't mean that they can be useful as models of some aspects of language processing behaviorally or in the brain. But I think all of these caveats need to be taken into account. For me, like we just talked about how language is separate from thinking. For us, for my group, I think these models are really,

great because they're kind of like a clean version of language. It's like language in a box. None of the other stuff, perceptual, motor, conceptual is in there, right? It's just linguistic statistics, which is what I think the human language system is maybe similar to, because it's separate from all this other stuff as well. And so we find that in the brain, these models actually capture neural responses to language quite well. But of course, if we want to build models, they can capture responses to language understanding in a deep sense, right? Like reasoning on representations that we get from language. So it's not just kind of getting some coarse approximation of meaning,

but actually doing some downstream reasoning on the things we get from linguistic input. Then I think we'll want to supplement these statistical language models with models of actual aspects of thought and reasoning, including social, social modeling of social like mental states. And there's exciting work happening in that direction at MIT and other places. So what else are you working on next? And what are the big questions that you still want to answer? Trying to use artificial neural network models as models of the mind and brain, I think it's a very promising direction. I think it's for the first time in the history of the kind of language research where we potentially can have

some kind of a close to an algorithmic level understanding of how language processing might happen. And we can go beyond putting words like syntax on some blob on your brain and just leaving it at that, which is where the state of the art has been for many years. But now we can actually build some fully implemented models that do something that at least in some aspects look similar to humans. Of course, the goal here for us as science does not to build a model that's as close to the brain as possible. although that is one goal because then it can also afford certain things. But I'm more excited to understand what it is about these models that leads them to capture something about human behavior or human neural response.

And so we're doing a lot of experimentation now on these models, messing with them in various ways because we have full access to them, right? We can lesion them. We can train them in different ways. We can mess with their architectures and try to see what are the properties that the model needs to have to provide a good fit to human cognitive or neural mental spaces. And I'm also very excited to try to understand how language develops.

So I'm involved in some collaborations. It's a very hard, pain-thinking effort to try to collect neural responses from young kids because most of the action is happening between, say, I don't know, like nine months and kind of two years. Like that's the sweet spot of development where language kind of comes to be. It still develops a little bit after that. really the bulk of it is there by like two and a half, three, like it's, you know, almost all there. And there are people here and in a few other groups trying to develop tools to be able to test younger kids in the scanner with silent sequences, really fun paradigms, all sorts of kind of ingenuity, experimental ingenuity.

And I think that could be another window to try to understand how language may relate to general reasoning abilities. as well as to kind of social perception, cognition, by looking at whether perhaps language shows, early language areas show more overlap with, for example, brain regions that support general kind of thinking and reasoning or areas that support some aspects of processing conspecifics, right, processing facial expressions, body language, thinking about others' intents and things like that. So hopefully in the next few decades we can make some headway. or maybe sooner. Well, those are big questions, and this is really great, great work that you're doing very interesting. Thank you so much for joining me today, Dr. Federenko.

That was really fun. Thank you so much. You can find previous episodes of Speaking of Psychology on our website at www.combecingofpsycher or on Apple, Stitcher, or wherever you got your podcast today. And if you like what you're here, leave us a review. If you have comments or ideas for future podcasts, you can email us at Speaking of Psychology at APA.org. Speaking of Psychology is produced by Lee Weinerman. Our sound editor is Chris Kondyen. Thank you for listening.

For the American Psychological Association, I'm Kim Mills.