1
00:00:00,000 --> 00:00:05,760
[Music]

2
00:00:05,760 --> 00:00:13,400
Welcome to the Language Neuroscience Podcast. This is Episode 39. I'm Stephen Wilson at the University of Queensland in Brisbane, Australia.

3
00:00:13,400 --> 00:00:21,800
A few episodes ago I spoke with Valentina Borghesani and Ryan Law from the Program Committee about the upcoming Society for Neurobiology of Language Annual Meeting in Geneva,

4
00:00:21,800 --> 00:00:24,400
that's coming up in September, October this year.

5
00:00:24,400 --> 00:00:33,200
In that episode we talked about the conference, the venue, the program, that kind of stuff, and if you haven't listened to it, I recommend going and checking that one out.

6
00:00:33,200 --> 00:00:38,000
But today we're doing a companion episode where we're really talking more about the science behind the conference.

7
00:00:38,000 --> 00:00:45,000
The best people to talk to about that are the scientists who are actually based in Geneva, who are part of the community that brought SNL to Geneva,

8
00:00:45,000 --> 00:00:49,600
and whose work reflects the intellectual theme of this particular meeting.

9
00:00:49,600 --> 00:00:56,000
I'm very pleased to be joined today by Nina Kazanina and Alexis Hervais-Adelman, both at the University of Geneva.

10
00:00:56,000 --> 00:00:57,800
Nina, Alexis, welcome.

11
00:00:57,800 --> 00:00:59,200
Hello, thank you.

12
00:00:59,200 --> 00:01:01,000
Hi, Stephen, thanks very much.

13
00:01:01,000 --> 00:01:08,800
It's great to meet you both, so Nina let's start with you. Can you tell me a little bit about yourself, your background, where you've come from, and what you're working on at the moment?

14
00:01:08,800 --> 00:01:14,400
I'm a cognitive neuroscience of language person, so a typical description I guess on your podcast. 

15
00:01:14,400 --> 00:01:34,000
I work at the University of Geneva now, and I've been here for the past two and a half years before that I was at Bristol University for about 16 years working the experimental psychology department, and the general theme of my work is

16
00:01:34,000 --> 00:01:45,200
cognitive neuroscience of language broadly construed with more and more emphasis on neural mechanisms that enable linguistic computation.

17
00:01:45,200 --> 00:01:49,800
Great, and what is the project that you're most excited about right now?

18
00:01:49,800 --> 00:02:01,600
Well, the SNL, if we're talking about generally, but if we're talking about research projects, it's probably what I've just said, which is

19
00:02:01,600 --> 00:02:14,000
anything related to neurobiological mechanisms underlying language and neurobiological mechanisms in its most, in their most precise form possible.

20
00:02:14,000 --> 00:02:25,000
Uh-huh, and do you have a particular empirical project you're working on that's kind of taking up your days and nights or like what's your sort of day to day at the moment?

21
00:02:25,000 --> 00:02:39,000
A lot of the stuff that we've been doing has to do with relations, so, uh, kind of representing relations within the semantic network and semantic memory network and why they're missing from there.

22
00:02:39,000 --> 00:02:51,400
How they can be inserted into there, so what does it take to represent our knowledge more fully than just a network of interrelated concepts as we do typically in a textbook?

23
00:02:51,400 --> 00:03:05,000
So, what does it mean to, um, what would it take to add the information on relations between concepts and not just have one unspecified association as a relation?

24
00:03:05,000 --> 00:03:11,400
Okay, so you're actually interested in like the details of compositionality, you're not like a just a bag of words type person.

25
00:03:11,400 --> 00:03:14,200
No, I'm definitely not a bag of words type of person. (Laughter)

26
00:03:14,200 --> 00:03:34,200
And, um, yes, I think it's, you know, generically one can describe the problem as, um, you know, take a typical representation of the, um, uh, semantic memory that's an interrelated network of concepts, cat is related to dog, cat is related to mouse.

27
00:03:34,200 --> 00:04:02,200
Nevertheless, the relations between those concepts are very different. What does it take to enrich the network to represent those relations? Some areas are absolutely not interested in the question because it's, you know, just the notion of association has been so over, uh, like sweet, uh, like such a sweeping notion and there is no need to kind of think about the details of relations, you know, that would be biologist sometimes psychology, but coming from the linguistic perspective, of course,

28
00:04:02,200 --> 00:04:18,200
knowing that cats chase dogs, but cats eat mice, you know, that's super important. And so, bringing in this information, which should be absolutely both trivial and fundamental becomes actually quite hard when you, when you try.

29
00:04:18,200 --> 00:04:22,200
Oh, yeah, absolutely. I get it.

30
00:04:22,200 --> 00:04:34,200
And Alexis, how about you? What's your background and where have you come from and what are you working on at the moment? All right. Yeah, I'm much like Nina. My background is really cognitive neuroscience of language.

31
00:04:34,200 --> 00:04:50,200
I started out in, maybe, in auditory processing with a particular interest in degraded speech perception, as applied to cochlear implant users, which is something that I'm very lucky to have managed to continue doing for the last 15 or so years.

32
00:04:50,200 --> 00:05:14,200
From there, I've sort of moved on as I shifted institution, I moved from the University of Cambridge to Geneva about 17 years ago, where I studied working on multilingualism and the context of simultaneous interpreting, which led to a really big expansion in my, my world of what I found exciting in the brain from the auditory system to subcortical areas, midline areas, the cerebellum.

33
00:05:14,200 --> 00:05:27,200
I then moved to the Max Planck Institute in Nijmegen for a couple of years before moving back to Switzerland to Zurich, where I spent six and a half years before moving back to Geneva about two and a half years ago.

34
00:05:27,200 --> 00:05:43,200
And my range of interests has now expanded to include difficult questions about syntax and composition and more lately, the fetal brain and the development of what I consider to be quite crucial to our language

35
00:05:43,200 --> 00:05:48,200
capacity, which is the audio-motor interface.

36
00:05:48,200 --> 00:05:55,200
I think that's it. Yeah, that's great. And there's a bit of an overlap with Nina there on the, on the compositionality part.

37
00:05:55,200 --> 00:05:59,200
And in some of your descriptions of your work, I saw the phrase extreme language.

38
00:05:59,200 --> 00:06:04,200
Can you tell us about like what does extreme language that you've worked on back in the day?

39
00:06:04,200 --> 00:06:11,200
Sure, I mean, by extreme language, sadly, I don't mean anything as exciting as profanity or taboo stuff.

40
00:06:11,200 --> 00:06:21,200
And rather the language system pushed to the limits. So, in the context of degraded speech perception, for instance, I think we can get a lot out of, we can enhance our knowledge

41
00:06:21,200 --> 00:06:31,200
a great deal by seeing the mechanisms that are implicated in processing speech when it is noisy, when it is spectrally degraded, when there are other challenges imposed on the system.

42
00:06:31,200 --> 00:06:50,200
It leads us to pose questions about what are the cognitive mechanisms are actually available to be engaged by our linguistic processes, which is the kind of thing that really started expanding my, my notions of what the language network is beyond the sort of simple left lateralized perisylvian brain areas.

43
00:06:50,200 --> 00:07:11,200
And another extreme is something that we characterize as extreme language control, which was simultaneous interpreting, which is a task where one has to juggle between two languages, which is something that a multilingual individual, in a typical community of context does not have to do. One typically chooses one language for momentary communication as a function of the needs of the interlocutor.

44
00:07:11,200 --> 00:07:26,200
Whereas a simultaneous interpreter is required to continuously access the source language of the speaker of a given speech and to convert that into a target language, which is a peculiar thing for the brain to have to do normally we switch between languages.

45
00:07:26,200 --> 00:07:30,200
Though not always, we can engage in code switching.

46
00:07:30,200 --> 00:07:37,200
And voila, sorry, the extreme is really there pushing the system to what I consider to be the unusual limits.

47
00:07:37,200 --> 00:07:49,200
Yeah, so you've got a very seminal paper on the you know using a fMRI to study simultaneous interpreters, which sadly we don't get into today, but like it's a, intriguing study.

48
00:07:49,200 --> 00:08:00,200
And you know, I think whenever I'm at SNL, I always love watching the sign language interpreters, you know, and especially like if the talk isn't good, I'll just focus on like watching the sign language interpreters and just thinking about what a feat that is.

49
00:08:00,200 --> 00:08:17,200
So, I'll just allow myself to trail that we have actually completed the acquisition of data on a study on sign language interpreting and FMRI and that will hopefully be, be pre-printed within the next month, so that we can all get a better view of how the brain of the sign language interpreters work.

50
00:08:17,200 --> 00:08:22,200
Oh, well, okay. Yeah, that's definitely worth keeping an eye out for. Cool.

51
00:08:22,200 --> 00:08:30,200
Okay, so now before we get into your science, let's talk a little bit about how we all ended up here, which is bringing SNL to Geneva.

52
00:08:30,200 --> 00:08:38,200
So, Nina, you submitted the original bid. Can you tell us about that? Like how did that come about? What was the vision that you had in mind when you wrote that proposal?

53
00:08:38,200 --> 00:08:51,200
So, the idea of hosting an SNL has always been in my head somehow, but it was not realizable until just a few years ago.

54
00:08:51,200 --> 00:09:01,200
I mean, until this recent years when I moved to Geneva where there is, finally I was in a situation where there was a big critical mass of people who were all interested in language.

55
00:09:01,200 --> 00:09:11,200
So that already kind of puts you out there thinking, okay, we can actually be a hub for, for, you know, an event or more than an event.

56
00:09:11,200 --> 00:09:20,200
And I think it was just, it just came naturally. We are part of a national competence research center.

57
00:09:20,200 --> 00:09:29,200
Actually, I should have said national center for competence of research NCCR, evolving language that was already mentioned by Valentina Borghesani.

58
00:09:29,200 --> 00:09:39,200
So, when you live in a context like this, the organizational side becomes much more, not necessarily trivial, but possible.

59
00:09:39,200 --> 00:09:47,200
And so, it was actually quite an easy thought to think, okay, one day I will host the big conference and the big conference for us does mean SNL.

60
00:09:47,200 --> 00:09:50,200
So, we just thought one day will host the SNL.

61
00:09:50,200 --> 00:10:10,200
And so, then the rest of it was about figuring out what the topic would. not the topic would be, but what would an ideal team, theme be and whether we can convince the organizing committee or the SNL board more correctly to play along with us, you know, in accepting this theme.

62
00:10:10,200 --> 00:10:16,200
But yes, it went smoothly, I think. And so there we are hosting it this year.

63
00:10:16,200 --> 00:10:23,200
And yes, and so what is, can you remind us for people that might not have listened to the previous episode, what is, what is the theme?

64
00:10:23,200 --> 00:10:39,200
Maybe theme is like actually a wrong description, but what I really meant was that it was the request to organize a top-down panel, which is a top-down panel on the neural architecture of language and other cognitive domains from an evolutionary perspective.

65
00:10:39,200 --> 00:10:51,200
So that's what I call the theme and that's what's important for us to bring in the evolutionary view and the across cognitive domains view into the SNL program.

66
00:10:51,200 --> 00:11:03,200
Right. So yeah, that's a panel discussion that's scheduled for the Thursday morning of the conference and is probably the part of the program where you're, you know, the unique vision of your center is most apparent, right?

67
00:11:03,200 --> 00:11:17,200
That's true. I mean, the title of that panel has a word evolutionary in it and the center is called NCCR evolving language. So, we can see that there is the same morphology happening in both parts.

68
00:11:17,200 --> 00:11:24,200
I think maybe equally important is one of the keynote speakers.

69
00:11:24,200 --> 00:11:45,200
And that's John O’Keefe who will be talking about his, you know, seminal work on spatial navigation and maybe we'll come back to this point, I mean, this point, as well as his less well known, but equally groundbreaking in my view, work on language.

70
00:11:45,200 --> 00:11:53,200
So, together with the panel, these are the two things that we think are Geneva specific and you know, have been our input to this year's conference.

71
00:11:53,200 --> 00:12:06,200
Yeah, definitely. Yeah, we're definitely going to circle back to him. And can you tell us who's on the panel and what's the vision behind the lineup that you guys chose?

72
00:12:06,200 --> 00:12:13,200
So, there are four people on the panel and the panel actually is also not a typical panel in the roles of those four people as we'll see.

73
00:12:13,200 --> 00:12:26,200
So, the four panelists, the three panelists are Sonja Kotz, who is a cognitive neuroscientist at the University of Maastricht.

74
00:12:26,200 --> 00:12:44,200
And Yannick Becker, who is also a cognitive neuroscience of language person at Max Planck University in Leipzig and Richard Hahnloser who is a neuroscientist and a computational neuroscientist at the University of Zurich and ETH.

75
00:12:44,200 --> 00:12:57,200
So, in addition to those three panelists, we have a moderator that's Paul Middlebrooks, who is a colleague of yours in the hat that you're having right now, wearing right now.

76
00:12:57,200 --> 00:13:13,200
He is a host of the brain inspired podcast that has had many interesting episodes, not necessarily on language on the, although some of them are on, you know, on language, but they 

77
00:13:13,200 --> 00:13:30,200
are for neuroscience generally so we thought that he'll be an amazing moderator to in the context like this where we aim to take the discussion outside of just the language domain and into the cognition more generally and into the evolutionary perspective, so 

78
00:13:30,200 --> 00:13:40,200
just, you know, going beyond what's the kind of normal scope and conversation for us and asking questions that would link the domains.

79
00:13:40,200 --> 00:13:49,200
But it's a really good idea to, you know, bring in an outside person to be the moderator. I think that'll be really fun and yeah, I can't wait for that conversation.

80
00:13:49,200 --> 00:13:57,200
You know, they span such a range of things they work on, right? Like primates, songbirds, humans, comparative and evolutionary.

81
00:13:57,200 --> 00:14:02,200
Like what are you kind of hoping that, that panel discussion surfaces for us?

82
00:14:02,200 --> 00:14:11,200
We really wanted to create a space for some public creative thinking or at least that was my personal take on it was that, at

83
00:14:11,200 --> 00:14:23,200
SNL we've seen some fantastic debates in the past. We've seen a lot of very collegial discussion. See, you know, very, very interesting reviews of the field, but not to say that we're trying to do something totally different, 

84
00:14:23,200 --> 00:14:32,200
but we felt that this was an opportunity to really bring in, we've obviously had animal research in SNL before notably enough 

85
00:14:32,200 --> 00:14:42,200
and we felt that was a really good precedent for bringing in some other species and to try to use that as inspiration for talking a bit more globally about the language faculty itself,

86
00:14:42,200 --> 00:14:49,200
and particularly with this context that Nina has elucidated of this evolving language research cluster that we have across Switzerland.

87
00:14:49,200 --> 00:15:04,200
To try to bring things together in a, as I say, a space that allows for some creative public thinking without people hopefully, without a panelist, without an audience, being too constrained by what they might normally feel is staying in their lane.

88
00:15:04,200 --> 00:15:09,200
I hope that we'll be able to foster that that sort of open and creative environment.

89
00:15:09,200 --> 00:15:28,200
Yeah, do you think that the field has been a little bit distracted by the human specificity of language that's been so emphasized, you know, by Chomsky and, you know, many of us that have drawn from his work. Do you think it's kind of gone a bit too far in that direction and needs to be pulled back?

90
00:15:28,200 --> 00:15:51,200
What a fantastic question. I say that with a smile because I'm not entirely sure how to address it. In many ways, I would say of course, and now and overly narrow focus on, on any aspect can, can lead to sort of slightly short sighted thinking, but we of course have to acknowledge the apparent human uniqueness of language.

91
00:15:51,200 --> 00:16:20,200
I think I've been heavily influenced over the last six, seven years in Switzerland by people looking at animal communication and that has that close proximity to them that has been fostered by specifically the NCCR evolving language really has led me to question some of my previous assumptions and to try to understand better how we can use data from non-human species to understand a little bit more about what the broader language network actually is to try to set things into a more.

92
00:16:20,200 --> 00:16:34,200
cognitive set of explanations rather than a human set of explanations and I think that's something that's, you know, common to, to many of us within this, this research enterprise. There are well over 100

93
00:16:34,200 --> 00:16:45,200
researchers in there and I think with slightly, building a slightly different vision. Again, not to claim that, that is in any way a particularly novel thing, but it's novel to me as an individual.

94
00:16:45,200 --> 00:17:06,200
I'll add to this, I think it was a useful period of let's call linguistics or you know cognitive science of language that was needed for linguists in order to figure out what are human specific side of

95
00:17:06,200 --> 00:17:16,200
what language is, like how languages differ, in which ways they are universal what are the

96
00:17:16,200 --> 00:17:34,200
things about language that we take as fundamental what is the core that we want to really talk about and then you know what is the core in in biological sense as well, but now that it has been you know done let's go some degree at least I think you can ask a

97
00:17:34,200 --> 00:17:48,200
different question.  You can ask the question of how that property of language evolved or capacity of language evolved and appeared in a human and this you know you can give the answer that it was you know a single mutation but I think that's,

98
00:17:48,200 --> 00:18:05,200
even if that were the answer you could still say okay it's single mutation but it's actually been added to all the other apparatus that we will really had at the time this mutation happened and if this is added you know it probably interacts in a way that's not trivial and not clean.

99
00:18:05,200 --> 00:18:25,200
So, even if you accept a very strong view of a single mutation on language you still need to say okay the single mutation happened on the background of you know some working memory representations or some you know linear order representations and so on so forth like the interaction is still needs to be explained and,

100
00:18:25,200 --> 00:18:45,200
and I think you know as a cognitive neuroscience field we probably are in a good time, I’m not sure whether we're in a good shape but where you know we have kind of a crude enough information and knowledge and are able to probably make an attempt to interface with other domains.

101
00:18:45,200 --> 00:19:02,200
Yeah. Great answers. Yeah, I mean I personally was never particularly wowed by the argument that it was a single mutation. I don't know that I have to go quite there, but like yeah that's even in best case scenario, right?  Like…

102
00:19:02,200 --> 00:19:29,200
So, yeah this this kind of leads very naturally into the paper of yours that I wanted to talk about today just to get a little bit deeper and it's called: ‘The neural ingredients for a language of thought are available’ and spoiler alert it's going to involve rodents as well so you definitely like bridging that you know it's not super human centric it's definitely thinking about human cognition from a very broad biological perspective and

103
00:19:29,200 --> 00:19:51,200
Co-authored with David Poeppel and it's in Trends in Cognitive Sciences, I think 2023. Is that right? That's correct. Okay, 2023. So, ‘The neural ingredients for a language of thought are available’. Can you tell us, it’s a bold title. Can you tell us first what is a language of thought and why is that important?

104
00:19:51,200 --> 00:20:20,200
Language of thought maybe it's better to use the less kind of conflictual term system of thought is the idea of orders that the, there is a kind of thought that does a lot of work for us as humans and that's let's call it propositional thoughts which is not a unique type necessarily, there might be iconic thoughts, there might be kind of non, like, yeah other types of thought but the type of thought that's called propositional

105
00:20:20,200 --> 00:20:49,200
is systematic, it has its compositionality and that compositionality is in some sense has nothing to do with language but at the same time has everything to do with language because it's very similar to it, okay? So, it's not about the statement that you're thinking using verbal descriptions and you know you're using language to think. It's not at all that. It's more that whatever, whenever you're producing complex thoughts, you are using

106
00:20:49,200 --> 00:21:04,200
ingredients and you're combining them together in the same way that you would you know take words and combining them and then get them into larger sentences. So, that's how I understand the main, one of the main claims of the system of thought by Fodor.

107
00:21:04,200 --> 00:21:10,200
Yeah, Jerry Fodor. Yeah, from the mid, probably dating back to the mid-seventies, I think he wrote the book on that.

108
00:21:10,200 --> 00:21:26,200
And why has the neuroscience community been skeptical that the brain could implement something like a language of thought as envisaged by Jerry Fodor?

109
00:21:26,200 --> 00:21:55,200
So, the language of thought system in Jerry Fodor’s take is highly symbolic. It requires taking symbols, being able to operate them with the like put them into a combination, swap them, replace them and so on so forth. So, it's a very highly symbolic system and there has been a great skepticism in the fact that symbols can exist in the brain. So, what we could find in the brain, we being neuroscience, I guess are

110
00:21:55,200 --> 00:22:23,200
are token, so you know, responses to a stimulus and if the stimulus changes then the response changes as well so like that that kind of you know not necessarily stimulus response idea from the behavioral perspective but more the inability to find anything that is more abstract than a response to a stimulus or perhaps inability is a wrong word actually in this context perhaps it's,

111
00:22:23,200 --> 00:22:52,200
in some sense lack of desire or lack of necessity because you know a lot of this experimentation is done in an animal context where you're showing stimuli which are just there you know in the moment and not necessarily the stimuli that are that you know, have the long term memory like words that we care about and so on. So, I think it's, it's basically a coincidence to a big degree that's you know what happened was that

112
00:22:52,200 --> 00:23:04,200
as of you know 2020, is the field of neuroscience thought that symbols are unavailable. Of course the other part of that debate is, or that claim comes from

113
00:23:04,200 --> 00:23:20,200
kind of the neural network route who are, who tends to be anti-symbolic and who tends to get the marvelous you know, outcomes that they do without necessity of without talking about symbols and

114
00:23:20,200 --> 00:23:39,200
therefore, the reductionist view of symbols. Right. So, the brain is just not obviously a symbolic processor. It's kind of some heavy and tight neural network and it's just not obviously has those kind of computational properties that you would need to underpin a language of thought is that a fair summary of what you're saying?

115
00:23:39,200 --> 00:23:45,200
That's correct. Yes. Like you don't need symbols. Symbols are overrated. They are, they are

116
00:23:45,200 --> 00:24:07,200
kind of a toy that was invented, you know, but that doesn't have physical reality and doesn't have necessity, like physical necessity. So, that you like symbols and you'd like to make the case that the brain does have the machinery to process symbols so what's the key move that you make on your paper to make that case?

117
00:24:07,200 --> 00:24:12,200
The key move is very simple. So, the key move is to realize that a

118
00:24:12,200 --> 00:24:24,200
basic type of symbol is for example a symbol that refers to a category. Okay? So, the symbol that represents the category of animals or fruit or whatever

119
00:24:24,200 --> 00:24:53,200
category might be interested in and so the key realization of that paper was that there is plenty of information like that that's available in rodents. So that, you know, that's the animals that we, you know, that we talked about because they are very well studied in the context of cognitive, of spatial navigation, which was the focus of that paper. Although, it didn't have to be because there are other animals studied in other context just maybe not to the same degree.

120
00:24:53,200 --> 00:25:22,200
But if I were to give you know the example that's kind of, that was one of the first things that's, that was the striking evidence of for symbols you know when we started talking about this with David, is the example of single cells that are found in rodents when they navigate and so there are like various types of cells found very, very often we talk about or hear about place cells, because they were the first ones found and these are the cells that respond to 

121
00:25:22,200 --> 00:25:40,200
a certain location so, or they fire when the animals in certain location in addition to those place cells, you know, they that were discovered in 1971 by John O’Keefe and I think Jonathan Dostrovsky,

122
00:25:40,200 --> 00:26:04,200
there are other cells that have been discovered since, among which are cells that are called boundary cells or border cells that respond to any object that represents an impediment of movement to, for to an animal. So, be it like a protruding, kind of boundary, right so, a wall in front of you or be it like a

123
00:26:04,200 --> 00:26:25,200
you know like a gap for example in the surface territory that the rat cannot cross or be it a water boundary, so something like a river. All of those things are caused the firing of a single type of cell which is called boundary cell and so the main

124
00:26:25,200 --> 00:26:54,200
claim in this paper was to look at the evidence on what exactly causes the firing of a boundary cell, realize that the firing that's caused, is caused by so many different objects and, and, objects of different kinds and, and, so just basically like extends the evidence that's there into a claim that the cell responds to an abstract notion of a boundary.

125
00:26:54,200 --> 00:27:04,200
That abstraction is so critical right? Like it's not, it's not just going to respond to one particular boundary that the rats learned about in this experimental environment.

126
00:27:04,200 --> 00:27:17,200
But you can manipulate the environment in different ways and put the boundary in different locations or turn the lights out or whatever and it's about the concept of a boundary not about any particular sensory representation of it, yeah?

127
00:27:17,200 --> 00:27:32,200
Exactly. It's like, it's exactly like you said. So, it's not about a boundary that's you know tall and wooden and smells you know in a certain way and has a certain color. It's about any impediment that might be

128
00:27:32,200 --> 00:28:01,200
present in the environment like be it, and if there is a new impediment, so if there is a new wall inserted into the environment and the boundary cell would fire to that new wall if there is, if the new wall is raised interestingly, so, it's still in the environment but no longer impediment to movement, the boundary cell stops firing. So, it's basically a cell that responds that has a meaning inside it so to speak and that's a metaphor by the way, the way I described this now.

129
00:28:01,200 --> 00:28:29,200
But the cell that has a meaning: I’ll fire every time there is an impediment to movement in a certain direction and just realizing how abstract that meaning of a cell might be or not might be, but how abstract that meaning of a cell is in the descriptions of the animal research, neuroscience research that is you know highly available and has been there for the last like 50, 60 years

130
00:28:29,200 --> 00:28:40,200
you realize that there is actually a very small gap between the notion, you know, those boundary cells found in rats and the notion of an abstract symbol boundary that we're assigned to humans.

131
00:28:40,200 --> 00:28:50,200
So, you're saying that single neurons in the rat’s hippocampus can do something like predicate logic as, as apparent in neurophysiological data?

132
00:28:50,200 --> 00:29:15,200
Exactly, so like you can then you know extend it one little step further. You can say that boundary is actually a predicate in human logic. So, it's a predicate that you know it's false or sorry it's true every time there is something that is impediment to movement and false otherwise and therefore you have a representation of a single place predicate like this in a rodent's brain as well.

133
00:29:15,200 --> 00:29:33,200
Right. It's a, it's a fascinating connection that you've drawn there. Not one that, I saw, saw John O'Keefe speak once before and I've totally wowed by it, the work. But I never really made the connection to, to predicate logic. So, that's really cool what you guys have done in this paper.

134
00:29:33,200 --> 00:29:42,200
The other key requirement would be not just that you have symbolic predicate but you're able to combine them right dynamically and productively.

135
00:29:42,200 --> 00:29:47,200
How would you build something like that on to the spatial navigation system?

136
00:29:47,200 --> 00:30:02,200
Well, there are some ingredients of that in the spatial navigation system as well. How far we can push them, that's a different question. But so, what's, what's known about spatial navigation system is that there are conjunctive cells there that take individual

137
00:30:02,200 --> 00:30:31,200
let's call the meanings or individual types of cells let's say head direction cell, that's, or signals that are captured by head direction cell. So, it's about whether the animal is facing west, north, east and so on so forth, and let's say the location of the animal in the space, so what's normally reflected by a place cell so whether the animal is located

138
00:30:31,200 --> 00:30:42,200
you know at a certain coordinates in a say the you know on a plane and so there are cells that take those two signals they had direction so whether they

139
00:30:42,200 --> 00:31:00,200
rat is facing north for example and the location and combine them together so that's well known and that happens for different types of signals so I've just given an example of you know place and head direction but you can also think of head direction and velocity or place in velocity or you know other combinations.

140
00:31:00,200 --> 00:31:29,200
And so, in that sense the existence of those conjunctive cells tells us that there is a way of putting together signals that are otherwise captured individually on its own by their own you know cell types. I'm slightly unsure or hesitant to say more than that and to make it stronger because of course when we talk about 

141
00:31:29,200 --> 00:31:41,200
compositionality and like putting things together in language really describe a very clean case of everything is combined with everything as long as grammar allows, right? So, any adjective can go with any

142
00:31:41,200 --> 00:31:52,200
noun and any noun can go with any verb so to speak and so on so forth unless semantic, until semantic restrictions come into place, right and I think in the

143
00:31:52,200 --> 00:32:16,200
animal neurobiology the claim of this kind of spontaneous kind of sweeping computation where you put anything next to anything and combine them together that's probably has not been, like the neurobiological side of this has not been shown to that same extent but some combinations that are

144
00:32:16,200 --> 00:32:29,200
flexible not to the ultimate degree but to a high degree have been shown and so this is like the conjunctive cells that I have yeah describe previously are those come like those pieces.

145
00:32:29,200 --> 00:32:42,200
So, yeah. You've, I mean you've shown conjunctivity but maybe you know recursion for instance might be a bit, a bigger ask from the, the rodent spatial navigation system but

146
00:32:42,200 --> 00:32:54,200
it's an intriguing, it's an intriguing start to see like what kind of abstractness that there is documented in mammalian brains.

147
00:32:54,200 --> 00:33:03,200
So, these cells we've been talking about like border cells and head direction cells, they’re mostly found in the hippocampus and surrounding regions, right? So,

148
00:33:03,200 --> 00:33:12,200
are you saying that that's like the primary locus of human thought or is the spatial system just one of the substrates that you think we could be building on?

149
00:33:12,200 --> 00:33:18,200
Thank you for asking because you know that's an important clarification that I should always make.

150
00:33:18,200 --> 00:33:42,200
David and I never meant to say that you know sort of hippocampal or you know the, or that thought is more related to spatial navigation than to say numerical cognition or to action or to any other cognitive domain that you could you know think about it's, it's just a proof of principle argument it's the reason for spatial navigation being such a good

151
00:33:42,200 --> 00:34:02,200
case for us is because it's been studied and it's been studied extensively and it's been studied in the manner that we can I think only and we admire so there is depths of research and there is depths of knowledge there but I'm sure that's or you know my personal view and I think David's personal view is that

152
00:34:02,200 --> 00:34:31,200
should we have information of the same level of detail about any other domain and like let's say you know action, numerical cognition, nothing comes else to mind, but I'm sure there are other domains, it's exactly the same thing. It's that the combinatorial principles and the architectural principles probably at their basis are very similar and then of course there are differences.

153
00:34:31,200 --> 00:34:48,200
Because in some domains you care about stuff that's not there, that's unimportant for other domains but these are you know, building like these are additions to the otherwise similar general kind of neurocognitive, neurobiological basis.

154
00:34:48,200 --> 00:35:04,200
So, not just the hippocampus, okay? Good. So, your TiCS paper is a perspective paper to theoretical paper.

155
00:35:04,200 --> 00:35:12,200
Have you thought about where you want to go with this next? Do you have like an empirical agenda to follow up on this work?

156
00:35:12,200 --> 00:35:40,200
We do and we don't. So, we do of course because basically yes the dream and the intention and the desire is to go down and be able to, by go down I mean that you be able to make and verify those claims in humans, right? And so, to see what it's, whether we, as humans we also have boundary cells

157
00:35:40,200 --> 00:35:52,200
and you know conjunctive cells and so on so forth. Some of those parts we already know actually so like some of the research by

158
00:35:52,200 --> 00:36:06,200
people like Rodrigo Quian Quiroga and others you know on concept cells actually give us ingredients of those same types but in humans. There are other things as well that can be added to this list or work by

159
00:36:06,200 --> 00:36:19,200
Andreas Nieder for example where he's working on numerical cognition and so he's looking for number cells in the human brain so there are several labs that are recording

160
00:36:19,200 --> 00:36:35,200
in humans this are always you know intracranial recordings all often from the hippocampus but not always and so like there we see the evidence for this, for this individual neurons that also have quite abstract representations.

161
00:36:35,200 --> 00:36:47,200
Going beyond that is harder because going beyond that and that's a super interesting and super necessary step in order to you know kind of make the next step is to look for,

162
00:36:47,200 --> 00:36:56,200
for the neural architecture and mechanisms that enable combination and that has been

163
00:36:56,200 --> 00:37:11,200
and I don't know of any evidence for this in humans so far, but you know one day hopefully it will be available and in terms of our own work you know so we are trying to

164
00:37:11,200 --> 00:37:19,200
with colleagues so that's the art measure van from the University of Geneva we're starting to record you know intracranially from

165
00:37:19,200 --> 00:37:27,200
Pre-epileptic surgery patients in the hippocampus we cannot necessarily

166
00:37:27,200 --> 00:37:47,200
design our experiments in this standard way where we have you know an experimental hypothesis that we can test because all of those recordings are of course you know somewhat let's call it opportunistic in terms of you know where you record from and what you can show to the patients and so forth but we're,

167
00:37:47,200 --> 00:38:02,200
we're trying and we're hoping that you know maybe one day will have a spark that will allow us to address the questions that are you know next step questions beyond recording you know, concept cells and beyond recording individual predicates.

168
00:38:02,200 --> 00:38:17,200
Yeah. Well that's great. I think single unit is where you need to be and, and we just have to address the challenges that comes with I guess if you want to if those are the questions that you're interested in.

169
00:38:17,200 --> 00:38:25,200
Yeah, single units you know give a big promise of course because that has been you know there we can

170
00:38:25,200 --> 00:38:33,200
kind of try and follow the steps of domains like, like spatial navigation where they have

171
00:38:33,200 --> 00:38:39,200
to a big degree the same questions as us. So, they have the questions of you know,

172
00:38:39,200 --> 00:38:57,200
who are my primitives so let's say in our case that maybe you know phonemes and, and morphemes and you know and words even and then they have the questions of how those primitives are combined to describe the behavior or to yield the behavior that

173
00:38:57,200 --> 00:39:14,200
observe in you know rodents or any other animal system so you know, how come the animal calculates the past, how come they know where they're located, how come they, you know, navigate in a new environment and so on so forth so.

174
00:39:14,200 --> 00:39:30,200
And that latter part is very similar to the questions of you know, how do we as humans you know put words together into a sentence, how we comprehend them, you know, how we figure out what that sentence means and so and so and so on and therefore

175
00:39:30,200 --> 00:39:40,200
the single like should we be able to record humans in a single cells kind of setting we know what to do we know who to follow in in terms of just research program.

176
00:39:40,200 --> 00:39:53,200
Equally of course, this day is neuroscience the complementary or you know competing approach depends on your position is to record as many neurons as possible on the view that

177
00:39:53,200 --> 00:40:09,200
there is like, rather than you know single cell computation being the computation that's kind of you know going to be the answer to, to our, to many of our interesting questions it's the computation that's made you know by the whole neural assembly and the neural dynamics of this computation

178
00:40:09,200 --> 00:40:35,200
that's you know, that's assembly wide is the, are the, are the course that we need to understand before we can understand anything about interesting behaviors and so that complementary approach is just kind of practically maybe a bit easier with humans because there you can record you know your whole head EEG and MEG and model

179
00:40:35,200 --> 00:40:50,200
you know, this like wide populations of neurons just because you know, just because their starting point is wide scale and kind of fits the technique and view and so to the extent that that view is

180
00:40:50,200 --> 00:41:13,200
like in this view is growing now and bringing more and more interesting insights. So, for example the idea of manifolds that are that reflects you know certain kind of sub parts of behavior so to speak or computation, so that view can be equally fruitful I think in the case of humans.

181
00:41:13,200 --> 00:41:19,200
Yeah. Okay. So, we'll attack it from both ends and just see where we, where we learn things. Right?

182
00:41:19,200 --> 00:41:33,200
Great so thanks for sharing about that paper which I really enjoyed reading and I want to turn to a paper that that you just put out Alexis that also has a very evolutionary kind of explanatory biological bit to it. 

183
00:41:33,200 --> 00:41:47,200
But it's a very different piece of the puzzle and this one came out in PLoS Biology 2025 and it's called: How did vocal communication come to dominate human language? A view from the womb.

184
00:41:47,200 --> 00:41:58,200
Let's start with an observation that you make that sort of seems obvious but actually isn't, which is that all human language is spoken unless there's a very good reason not to.

185
00:41:58,200 --> 00:42:06,200
Like for instance sign languages in deaf communities so what is that exception tell us about the puzzle that we're faced with here?

186
00:42:06,200 --> 00:42:14,200
So, the exception that you allude to which is that sign language has spontaneously emerged in communities with a high incidence of congenital deafness,

187
00:42:14,200 --> 00:42:27,200
it tells us that language is not bound to a particular modality, and I think that's a really important thing to acknowledge whenever we talk about language is something I emphasize in my teaching that language is not equivalent to speech,

188
00:42:27,200 --> 00:42:56,200
and I think it does of course tell us something if we take it a more sort of sociological perspective the human impulse to communicate to build community and so on this sort of idea that we are fundamentally cooperative species that needs a way to actually interact with our conspecifics and therefore our language faculty is something that potentially enables this or the very least facilitates it in the way that we are used to and permits us to transmit information that needs to potentially

189
00:42:56,200 --> 00:43:11,200
more efficient ways of being together I think that's one perspective and I try not to forget that although of course the paper that you that you cite is very much about trying to solve the question of why do we speak in the sense of how come

190
00:43:11,200 --> 00:43:29,200
the vocal mode has turned out to be dominant and of course there are many, many arguments about you know the gesture first hypothesis or the coevolution of gestural and oral expressive communication and what we wanted to do with this paper was simply

191
00:43:29,200 --> 00:43:37,200
Sorry, can you tell us about that? People might not be down with the core biolysis oeuvre. What’s, what's the gesture first hypothesis?

192
00:43:37,200 --> 00:43:50,200
As a sort of sidebar theory, I’d rather not get too much into the controversy of the argument because I'm not an expert on this and I think it’s inappropriate to say too much about it. 

193
00:43:50,200 --> 00:43:57,200
There's a hypothesis that is very present in our community which is that the

194
00:43:57,200 --> 00:44:20,200
proto-humans had a gestural mode of communication which was somewhat effective and was the dominant mode of information transmission between individuals which then shifted to a vocal mode, and I find that some of the arguments that are put forward are not especially compelling to me as I tend to find them teleological in nature.

195
00:44:20,200 --> 00:44:38,200
A lot of them are articulated or feel like they're articulated along the sense of given that speech freeze the hands it was a very good idea to learn to speak because then you can do things like describe how you're making a tool while demonstrating the tool.

196
00:44:38,200 --> 00:45:00,200
There are other arguments that are articulated along the potential superiority of the vocal mode for communication at a distance potentially yes of course it might be again giving the idea that the sort of I'm anthropomorphizing even which is not a sensible thing to do that it would be a good idea to call

197
00:45:00,200 --> 00:45:14,200
at a distance that that might be something that is useful because you can alert your conspecifics to danger or you can indicate something interesting or necessary that is happening close to you and far from them.

198
00:45:14,200 --> 00:45:28,200
That's something that's an interesting intermediate stage because we understand very well that sort of emotional vocalizations which don't necessarily have linguistic content or something that exists of course throughout

199
00:45:28,200 --> 00:45:41,200
our primate cousins as well as through many many species and so that sort of intermediate stage is interesting but the idea that it would be a good idea to shout your thoughts across the

200
00:45:41,200 --> 00:45:57,200
prairie or at a great distance it might make you very very obvious to potential predators for example not necessarily the best idea. There's also an argument articulated around you can't use sign language in the dark or gesture is not very useful in the dark.

201
00:45:57,200 --> 00:46:12,200
Although this is of course true, humans are not a nocturnal species. We might not have had many opportunities for communicating in the dark anyway because potentially we've been asleep. (Laughter)

202
00:46:12,200 --> 00:46:23,200
These are not strong strong arguments, but they are worth thinking about because all of the, not all, many of the tales we tell when we're thinking about evolutionary arguments are just side stories and I think it's,

203
00:46:23,200 --> 00:46:36,200
It behooves us to think about the you know the contrary just so stories as well when doing this even just as an intellectual exercise if potentially a superficial and slightly silly one.

204
00:46:36,200 --> 00:46:51,200
Nevertheless, I think the question of how it has arisen remains open and what we wanted to do in this paper is simply put and put a little bit of focus on to

205
00:46:51,200 --> 00:47:14,200
something that we felt was, was ignored, which is that in terms of our ontogeny, we hear much earlier than we see or at least we experience the auditory world much earlier in our developmental trajectory than we see anything about the world and that's primarily because the womb itself is penetrant to sound.

206
00:47:14,200 --> 00:47:37,200
All be it filters sound very heavily between one kilohertz and ten kilohertz and of course there is a huge acoustic background of maternal noises: digestion or rich was all of the and whatever respiratory and cardiovascular noises nevertheless the fetal auditory system response to sound from roughly the 24th to 26th week of gestation.

207
00:47:37,200 --> 00:47:50,200
And its responses to sound have been shown to be increasing the complex and discriminatory over the gestation, over the last trimester of gestation. So yeah, so that's your,

208
00:47:50,200 --> 00:48:06,200
your fundamental answer as to why language is vocal and not signed is that the auditory system is basically available, develops and it has access to language much early in the visual system yeah?

209
00:48:06,200 --> 00:48:09,200
I think that's, that's one potential

210
00:48:09,200 --> 00:48:28,200
plank or maybe it's less than a plank and it's maybe just a splinter in the argument of why you know incrementally they could have been a shift towards that mode, and this thought this reflection was sparked by observations which they're not uncontroversial, but they are out there they've been published over the last 15 years or so primarily

211
00:48:28,200 --> 00:48:46,200
I'm anything from Kathleen Wermke’s group where there's a seminal publication by Mampe and colleagues in PNAS (2009), showing that newborn infants appear to have an accent in their very early cries which is influenced by the language of their gestational environment. Well, before we talk about, sorry

212
00:48:46,200 --> 00:49:06,200
Yeah, sorry, could we just, logically prior to the mode like the auditory mode of learning is the, the evidence that they are in fact listening and learning in the womb like, can you just run that by us first before you take us into the you know into the cutting edge of

213
00:49:06,200 --> 00:49:13,200
the ideas? Yeah, I mean, I'll try and keep it brief though because I could go over excited. (Laughter)

214
00:49:13,200 --> 00:49:30,200
And I think some for me some of the most you know the compelling there are many compelling pieces of this puzzle vis-à-vis in utero learning of auditory information and one of them is the well-known fact that a newborn infant preferentially

215
00:49:30,200 --> 00:49:54,200
to the voice of the mother more so than to any other female voice or to the voice of the father suggesting that the dominant maternal vocal input throughout gestation has had some kind of impact on tuning the infant that that preference can develop indicates that there's discrimination that there's some kind of learning and learning of some of the regularities and the acoustic properties of the mother's voice.

216
00:49:54,200 --> 00:50:14,200
There's also very interesting evidence from the 1980s from Hepper who looked at the newborn babies of mothers who watched the TV show neighbors twice a day versus those who did not watch the show twice a day. Were they Australian?

217
00:50:14,200 --> 00:50:27,200
They were British. Neighbors is very popular in the UK. I think it was actually yeah, or it could have been Ireland. I'm sorry, I really need to fact-check myself. It was definitely United Kingdom.

218
00:50:27,200 --> 00:50:46,200
But it is an Australian show indeed. I'm sorry give credit where credit is due. (Laughter) with a very distinct the very distinct sound and with the title music which is nine people of my generation but presumably not to the youngsters.

219
00:50:46,200 --> 00:51:05,200
I turned up the newborns who have been exposed to this to this show in utero responded very differently in terms of their behavior say movement and facial expressions compared to newborns who had not been exposed to this sound in utero. I call it sound, it’s music because it is a theme tune.

220
00:51:05,200 --> 00:51:34,200
This same study was followed up by some work using ultra sounds where they equally demonstrated that not only are the different behavioral responses postpartum in infants that in utero infants respond differently to this theme tune if it's played transabdominally if they know it. I say know, because I believe that the data suggests they do in fact know it or at least have encoded it as

221
00:51:34,200 --> 00:51:52,200
something that is regular and familiar. This although it sounds a little bit you know far out this kind of effect has been replicated multiple times and there's strong evidence for auditory learning in utero and I think this is not something that we

222
00:51:52,200 --> 00:52:17,200
necessarily should find surprising if a perceptual system works and receives input, we can I think logically expect that you know statistical and Bayesian learning is likely and Hebbian learning are likely to take place, there will be an intake of information that will lead to a tuning of the system and if we take it to the very very.

223
00:52:17,200 --> 00:52:46,200
primordial biological level if you stimulate any sensory system but for the sake of argument the auditory system the neurons that fire will cause a cascade of chemical change to happen and some of those changes are potentially or very likely also epigenetic in nature there could be the building of networks that is induced simply, I say simply, as a result of continued exposure and functional neural responses to sound.

224
00:52:46,200 --> 00:53:01,200
And this to me is something that we ought to explore in the domain of trying to understand the neural architecture of language as it is currently implemented in modern humans.

225
00:53:01,200 --> 00:53:20,200
Yeah and like you know you talk in your paper about how you know this audit, you know, sort of in utero virtual learning is certainly not unique to humans. You've actually got a lot of examples of different animals that do it for very good adaptive purposes and I was wondering if you could talk about the fairy wren example, I think that's a really striking example of

226
00:53:20,200 --> 00:53:49,200
useful in utero auditory learning. Yeah. So, the superb fairy wren songbird appears to learn in ovo from the parents what's known as a password, so vocal password, which it will have to reproduce in order to receive food from parents once it has hatched and this is a fantastic way of avoiding having hatchling from another species that gets

227
00:53:49,200 --> 00:54:03,200
fed by your parents. You call out and say I’m me and I belong to you and this has been demonstrated quite robustly and this is an example of something that is learned by an organism prior to its

228
00:54:03,200 --> 00:54:18,200
total exposure to the world. So, obviously the shell of an egg is permeable to multiple kinds of stimuli including auditory stimuli and it appears that the embryonic chick is able to acquire

229
00:54:18,200 --> 00:54:34,200
Information from the auditory domain and to learn it and to be able to reproduce it later on for a very important biological end of survival. Yeah, well if it's really important they're going to have to move to two factor authentication because you know (Laughter)

230
00:54:34,200 --> 00:54:41,200
Password is just not enough anymore apparently, according to my indeed my university administrators at least. Indeed.

231
00:54:41,200 --> 00:54:53,200
Maybe develop some olfactory clicks as well. I don't know. It's a very, it's very cool observations that you have in the paper about all the different sort of biological

232
00:54:53,200 --> 00:55:10,200
adaptations from this and then you talk about how we can learn more about this with in- utero magnetoencephalography (fMEG) or functional MRI (fMRI), that seemed very exciting to me. Can you tell me about like recent

233
00:55:10,200 --> 00:55:18,200
technological developments that make that possible? Sure. There are actually some older technological developments that

234
00:55:18,200 --> 00:55:31,200
they are somewhat forgotten. So, when getting into this, I was looking at what's out there in the field and I stumbled across a paper from I think it's as long as 1998 doing a demonstration of

235
00:55:31,200 --> 00:55:44,200
fetal auditory responses in-utero from a group in Nottingham which was just a proof of principle that this was possible in fMRI, and I was really surprised because all the work that I knew was much more modern the field of

236
00:55:44,200 --> 00:56:00,200
fetal neuroimaging using MRI is something that has really taken off over let's say the last decade and there are some pioneers in the field lately Moriah Thomason in the US who's scanned hundreds of fetuses there's also

237
00:56:00,200 --> 00:56:15,200
and now ended project European project that was centered around King's College London which is a developing human connectome project these have been really pioneering things that have collected quite large databases of fetal

238
00:56:15,200 --> 00:56:27,200
brain structure and resting state function. Very little has been done looking at I think to call it task MRI is a bit too far but stimulus induced brain responses.

239
00:56:27,200 --> 00:56:40,200
But where it has been done successfully is a couple of papers in the auditory domain one of which I think is worth highlighting because it was such a brilliant design which was to ask the mother to sing

240
00:56:40,200 --> 00:56:54,200
in the scanner and to look at the fetal brain responses to the internally transmitted song which showed reasonably reliable auditory cortical activation in a number of the fetuses that were scanned.

241
00:56:54,200 --> 00:57:00,200
In terms of the technological… What were they singing? Was it the theme tune from neighbors?

242
00:57:00,200 --> 00:57:12,200
If only. I think it was the lullaby that they typically, that they knew. I'm not sure if it was individualized or if they were ordered to sing a specific one. I think it was a personal favorite.

243
00:57:12,200 --> 00:57:14,200
Okay so you were saying.

244
00:57:14,200 --> 00:57:43,200
I have one of the, so technologically I don't know that there's you know there's general improvements in the quality and the speed of acquisition with fMRI that applies be it to adults all fetuses across any, any acquisitions. What I think is then one of the big developments over the last decade or so in addition to people pioneering these big database approaches has been a generalized acceptance that it is not unsafe to put a pregnant individual into an MRI scanner.

245
00:57:43,200 --> 00:57:54,200
Even though our consent forms obviously seek to needlessly to avoid needlessly exposing any pregnant person to whatever intervention be it MRI or anything else.

246
00:57:54,200 --> 00:58:12,200
We understand now from an enormous amount of pre-clinical work that's been done in MRI that there are no reports that I know of and so far that have been clearly reported of any negative outcomes for a newborn who has been exposed to MRI in-utero.

247
00:58:12,200 --> 00:58:40,200
And one of the major fields where this is you know been able to be tested has been in pioneering work on pre-surgical work up for in-utero spina bifida surgery which is something where MRI images of a fetus have been generated in order to allow a surgeon to prepare to actually intervene directly in-utero. All of the by now thousands of cases of this

248
00:58:40,200 --> 00:59:01,200
have not shown any particularly notable outcomes that concern for example hearing impairment in a newborn that might be one of the fears that we could have if we expose a fetus to loud noise for example. Right. So, to the best of our knowledge the safety profile is increasingly well accepted.

249
00:59:01,200 --> 00:59:27,200
Certainly it's become accepted by ethics committees that for small scale studies, it is, the cost benefit analysis appears to be reasonable for the people evaluating our research and I'm confident that you know we are working within a safety profile that is acceptable and this is something that's only happened because they have been now so many scans that have enabled us to build those, those safety profiles so

250
00:59:27,200 --> 00:59:38,200
I think that's one of the critical developments is really a generalized acceptance that it is it is not fundamentally dangerous or potentially dangerous to them.

251
00:59:38,200 --> 00:59:52,200
I can't wait to see our what we are going to learn about the emerging auditory responses to what you call task MRI, and I like the idea of like a fetus doing a task but yeah I know what you mean. We've wondered.

252
00:59:52,200 --> 01:00:02,200
Yeah. I mean we can't we obviously can't elicit many responses is why we're looking at the brain as a covert indicator of what's happening.

253
01:00:02,200 --> 01:00:11,200
So, you were going to mention earlier that you're also, that I kind of pulled you back to talk about auditory first but I know you wanted to talk about

254
01:00:11,200 --> 01:00:23,200
audio motor learning that happens early and I and I, my sense from your paper is that this is much more speculative and cutting edge than the then the strongly established evidence for

255
01:00:23,200 --> 01:00:30,200
prenatal auditory learning but what do you think about auditory motor learning that happens early on?

256
01:00:30,200 --> 01:00:41,200
So yeah, I mean, one of the things that inspired this research was the work of Kathleen Wermke, who you have repeatedly demonstrated that newborn infants seem to

257
01:00:41,200 --> 01:00:53,200
Apply an accent, a pitch accent to their very, very early cries and it is possible that they acquire this accent somehow in-utero.

258
01:00:53,200 --> 01:01:21,200
And this is a fascinating phenomenon because it's not possible to practice articulating in-utero to develop your accent. So, it implies that there might be, might be being a strong underlying contention here, a mechanism for mapping things we hear onto action patterns in some way. At the very least it suggests that there's a way of extracting some kind of pitch accent related regularity

259
01:01:21,200 --> 01:01:30,200
into some form that is stored and can then be applied to the cry which is a reflexive thing so it's quite a potent,

260
01:01:30,200 --> 01:01:36,200
 it was something that is applied to a you know, prepotent response so it must be a very strong

261
01:01:36,200 --> 01:01:47,200
connection between whatever is abstracted onto the production of the cries now the evidence for the accents is not on controversial and there is discussion about whether this exists.

262
01:01:47,200 --> 01:01:59,200
But it remains if it is a thing, it remains a very intriguing part of this particularly because there's a general argument about what is the function of crying what does it serve the infant.

263
01:01:59,200 --> 01:02:11,200
And of course, there's a whole slew of work on adult brain responses to cries and how hearing cries maybe activates mechanisms that are to do with preparing to give care to an infant.

264
01:02:11,200 --> 01:02:37,200
And so one of the things that we thought about is maybe an accent that signals you is having an acoustic similarity to your group might be something that tunes you as a successful infant it might be a way of optimizing your opportunities to get good care give attention from potential care giver, as in a way that an infant that produces a less well matched accent might not.

265
01:02:37,200 --> 01:02:53,200
And we're currently investigating whether there is any evidence of adult brain responses being different as a function of an infant cry accent whether it matches the native adult’s native accent or not and to see if that

266
01:02:53,200 --> 01:02:57,200
for example, has any influence on limbic system engagement.

267
01:02:57,200 --> 01:03:21,200
And I think this is important in the evolution perspective because obviously fostering a communication mode that requires tuning to the culture rather than being entirely arbitrary and particularly doing so, so precociously would also potentially be something that adds to this dominance or adds to the reason why the vocal mode dominates.

268
01:03:21,200 --> 01:03:32,200
And again, the vocal mode is not the only mode for successful communication sign languages are fully formed and fantastic languages that can express everything that that all languages can too. So, there's no

269
01:03:32,200 --> 01:03:42,200
you know, functional bottleneck why sign languages wouldn't dominate there's just we put the accent on why as being well there's just this slight

270
01:03:42,200 --> 01:03:57,200
that vorlauferform we say in German and there's just this little precursor that's just a bit earlier. Yeah, I mean some people like it, that’s why the left hemisphere is dominant for language. It’s just because, when I say some people, this goes all the way back to Broca, 

271
01:03:57,200 --> 01:04:11,200
pure speculation that if it just if it just develops even slightly earlier than that will just be the more natural place for language to emerge and we obviously know that language can emerge in the right

272
01:04:11,200 --> 01:04:15,200
It just refers not to, and it could just be a very small

273
01:04:15,200 --> 01:04:43,200
advantage that causes it to take one path rather than the other. So Nina turning back to you like do you see connections between your work and Alexis work? I mean you're in the same evolving language group. I can't reproduce the Swiss acronym for it, but I like, what kind of commonalities are there between you guys and between in the in the wider group that you're a part of?

274
01:04:43,200 --> 01:05:05,200
Yes, so probably the kind of things you've heard today are not the best examples of where Alexis and I connect the in the closest possible way but you know if we had more time on the program, we for the, you know, made that long connection. But for the sake of a shorter answer and a broader answer in the context of the NCCR evolving language,

275
01:05:05,200 --> 01:05:17,200
so, I think one thing that unites us all, is the interest in the evolutionary aspect and you know the word evolving and evolving language is very broadly constructed was,

276
01:05:17,200 --> 01:05:30,200
it's meant to represent you know the standard language evolution in the sense of you know how language evolved across species and also across time in terms of macro linguistic evolution of language families and so on.

277
01:05:30,200 --> 01:05:58,200
But also, it's taken to mean you know this evolving reflects the evolution that happens within the time span of an individual beats healthy timespan or a disease timespan. So, at that very glow or sorry like I just I'll just add one more bit and whether like healthy you know clinical timespan or

278
01:05:58,200 --> 01:06:17,200
evolution in terms of going from one language to more than one language so from the situation of say monolingualism to multilingualism within a person and so within that very global view of course everything that you've heard from any one of us from either of us

279
01:06:17,200 --> 01:06:37,200
like very much fits into this global view where we're all trying to inform each other and you know inspire each other and so on. More directly, you know, like there Alexis and I talk a lot you know, the last conversation was yesterday, we were discussing a project on actually cochlear implants, and the

280
01:06:37,200 --> 01:06:53,200
possibility of using a very simple and hopefully effective paradigm that's called frequency tagging to tap into the

281
01:06:53,200 --> 01:07:13,200
perception of people with cochlear implants more specifically you know whether they hear you know after being implanted with a cochlear implant like whether they hear acoustic input as you know one single percept or whether it doubles and that's

282
01:07:13,200 --> 01:07:31,200
a request that you know or a point that we're discussing with the audiologist colleagues of ours and we're just trying to make your practical steps on it from our side as and develop ways of testing this so that's you know that's a very short answer on a type of project that Alexis and I could

283
01:07:31,200 --> 01:07:46,200
collaborate on but there are many others, and you know like it's you know having the word now I'll say that you know it's great to have him as a colleague somehow reciprocates this opinion. Absolutely. Thank you very much. 

284
01:07:46,200 --> 01:07:49,200
Do you reciprocate?

285
01:07:49,200 --> 01:08:18,200
I very much do. I think it's, as Nina pointed out there, we have so many long and very stimulating conversations and it's a, it's genuinely a hugely mind expanding to have been exposed to Nina's view. I mean everything that she described from the paper that you cited earlier is something that prior to my you know, my personal interactions with her was very very distant from my, my mind but now is, is really present and I think it helps to, you know have a colleague

286
01:08:18,200 --> 01:08:32,200
who frames things so robustly and precisely in a very you know a very different format to what I do, and I think the complementarity is, is evident and it really is fantastic. Yeah. I'm really looking at 8 o’clock in the morning though.

287
01:08:32,200 --> 01:08:42,200
8 o’clock in the morning might be early for your opinions but in Brisbane that would be like you know lunchtime.

288
01:08:42,200 --> 01:09:01,200
So, we're nearly out of time. Before I let you go, well we're probably right out of time but in any case you know I know from having worked on organizing the conference in a past year how much work it is and I know you guys have probably put in a normal amount of work and making this conference happen.

289
01:09:01,200 --> 01:09:04,200
And what are you both most looking forward to about it?

290
01:09:04,200 --> 01:09:16,200
I hope that there will be you know one or more than one person who comes to this conference expecting whatever they might expect and leave the conference thinking wow.

291
01:09:16,200 --> 01:09:29,200
Like there was something that caught my attention in the way that I could not expect and in the way that will you know transform my thinking and I think if that happens that will be an amazing outcome for me.

292
01:09:29,200 --> 01:09:39,200
We haven't had time to talk about you know the panel and the keynote speaker one of the keynote speakers, John O’Keefe, but I hope that you know,

293
01:09:39,200 --> 01:09:58,200
like I'm certain maybe it's too strong to say but I'm really strongly hopeful that, you know, either of those occasions and many others of course that are you know beyond our planning and kind of outside of our planning but equally present will be those opportunities.

294
01:09:58,200 --> 01:10:22,200
Yeah, well we didn't, that's, that's a great hope and we didn't, we didn't talk about specifically what he's going to talk about in his keynote and maybe it's good to like you know, keep, keep it as a surprise for everybody. But you know obviously you know, you, you, you built your paper that we talked about is like constructed on findings that he was the generator of you know

295
01:10:22,200 --> 01:10:31,200
so many decades ago, now he's going to be at the conference and you and you mentioned that he's got work on language as well, so I'm really intrigued to see what that's going to be all about.

296
01:10:31,200 --> 01:10:39,200
So yes, let's keep it as a secret, but I think you know there is it's a it's a secret that's worth discovering.

297
01:10:39,200 --> 01:10:42,200
Yeah, and how about you Alexis?

298
01:10:42,200 --> 01:10:59,200
I'm excited to fill Geneva with our community and see all of the spontaneous and maybe less spontaneous interactions that arise from that in the you know various wonderful venues that we have for hosting people.

299
01:10:59,200 --> 01:11:21,200
And as Nina I'm really excited about the prospect of people coming and being exposed to something that maybe they hadn't ever planned to listen to. I think, it's one of the great strengths of SNL is this single track format that means that we're all in it together we hear about things that we might not necessarily have chosen to in a in a parallel session.

300
01:11:21,200 --> 01:11:40,200
And there's so much scope for picking up new information from a subtly different field that's going to inspire hopefully productive collaborations and also for the early career researchers to get such a broad overview and to see where you know where this takes them when they are PIs later on.

301
01:11:40,200 --> 01:11:51,200
I think we're it's formative and I'm excited to be able to contribute to that the Geneva might not be more so than any other SNL but we're hoping to do our best to make sure to contribute to that that great tradition.

302
01:11:51,200 --> 01:11:55,200
Oh, I'm sure you will I'm sure it will be great I can't wait.

303
01:11:55,200 --> 01:12:02,200
Well thank you both so much for your time today, it's been really fun. I'm getting to know more about your work and chatting to you both and getting to meet you both.

304
01:12:02,200 --> 01:12:34,200
It's been a pleasure. Thank you, Stephen. Alright. So, that is it for episode 39. Thank you, Nina and Alexis, for a great conversation. If you want to follow up anything we have discussed, I am going to link both of those papers in the show notes and at langneurosci.org/podcast and regarding the conference that we have been encouraging you all to go to, you can learn more about that at neurolang.org. Thank you to Marcia Petyt for transcribing this episode and thank you all for listening. See you next time.

306
01:12:34,200 --> 01:12:39,200
[Music]