WEBVTT

00:00:03.477 --> 00:00:10.497
This is the Convergent Science Network podcast. Leading researchers in the domain

00:00:10.497 --> 00:00:16.737
of neuroscience, brain theory and technology are interviewed by Paul Verschure and Tony Prescott.

00:00:19.897 --> 00:00:23.697
Paul Verschure here for the Convergent Science Network podcast,

00:00:24.677 --> 00:00:28.757
together with Tony Prescott. Scott, and today we're talking to Barbara Findlay,

00:00:28.877 --> 00:00:31.777
speaker at our BCPD Summer School 2015.

00:00:32.937 --> 00:00:38.557
So Barbara, you've been very focused on understanding brain evolution.

00:00:40.697 --> 00:00:46.897
You emphasize issues of segmentation of brains, commonalities between different brains.

00:00:48.997 --> 00:00:54.257
What do you see as the underlying principles of brain evolution?

00:00:57.577 --> 00:01:00.537
Um, a very large question to start with.

00:01:00.657 --> 00:01:08.837
Okay. So, um, I think, I think the underlying question that I've come down on is the one that I, uh,

00:01:09.037 --> 00:01:15.737
that emerges from a lot of research in evolution and development at Evo Devo

00:01:15.737 --> 00:01:22.157
field, which is how do you design something that is adaptable?

00:01:22.157 --> 00:01:27.917
That is responds to change and robust does not respond to change simultaneously.

00:01:30.277 --> 00:01:39.597
So that's how I've started characterizing how to best understand how nervous systems evolve.

00:01:39.717 --> 00:01:42.357
That is distinguished in my mind from the,

00:01:43.808 --> 00:01:50.688
The kind of random walk model of evolution that people have had to a large extent

00:01:50.688 --> 00:01:53.588
in behavioral biology so far,

00:01:53.708 --> 00:02:02.388
which is to imagine a brain and an organism as a series of ad hoc evolved patches

00:02:02.388 --> 00:02:10.188
as opposed to a set of theme and variations on this adaptation and robustness.

00:02:11.288 --> 00:02:15.368
So is this what you mean when you talk about the filter idea?

00:02:15.368 --> 00:02:19.648
Idea yeah so um if

00:02:19.648 --> 00:02:22.708
you think about um evolution of the

00:02:22.708 --> 00:02:26.048
brain only in terms of adaptation and maximizing say

00:02:26.048 --> 00:02:32.088
how well you can detect a particular set of flowers or evade a certain predator

00:02:32.088 --> 00:02:37.528
or something like that you you come up with um you know particular kind of of

00:02:37.528 --> 00:02:43.688
nervous system that you imagine is being optimized for all these very specific uh kinds of um.

00:02:45.368 --> 00:02:46.488
Kinds of behavioral problems.

00:02:46.648 --> 00:02:50.428
But if you imagine that the same nervous system has to,

00:02:51.188 --> 00:02:59.748
also survive horrible catastrophes and shifts of niche and all those sorts of

00:02:59.748 --> 00:03:02.168
things as also part of its repertoire,

00:03:02.408 --> 00:03:05.588
you come up with a different kind of nervous system that you need to evolve

00:03:05.588 --> 00:03:09.588
rather than this sequentially adding of components.

00:03:09.988 --> 00:03:15.608
You come up with something that has to have a certain set of sort of core survival

00:03:15.608 --> 00:03:21.268
recognition learning functions as part of its repertoire, not only a set of,

00:03:21.268 --> 00:03:22.868
you know, maximum adaptations.

00:03:23.308 --> 00:03:28.088
So is it, then you're thinking less about evolution on the species level and

00:03:28.088 --> 00:03:29.988
more about phylogenetic change?

00:03:30.428 --> 00:03:36.768
It pushes you in that direction. I mean, all the causal structure is going,

00:03:36.968 --> 00:03:42.028
the causes, is eventually we're going to come down to the survival of this versus that individual,

00:03:42.688 --> 00:03:46.428
that I think the idea is to think of.

00:03:47.688 --> 00:03:56.248
Cumulatively over individuals, both the sort of adaptive maximization kind of

00:03:56.248 --> 00:03:58.908
things, can I recognize that particular kind of spot?

00:03:59.028 --> 00:04:04.908
Or, you know, can I find this kind of food the best versus can I find any food?

00:04:05.088 --> 00:04:06.988
Can I get out of this hole?

00:04:07.568 --> 00:04:13.028
So we've got two kinds of individual histories to amass, not just the adaptation

00:04:13.028 --> 00:04:18.908
one, but also the, you know, dealing with any eventuality one as well.

00:04:19.328 --> 00:04:23.128
And so the ones that the runs remain alive have to do both.

00:04:23.368 --> 00:04:27.468
So that would be the fitness aspect of it, like how appropriate is the behavior

00:04:27.468 --> 00:04:29.028
coming out of that system in the end.

00:04:29.468 --> 00:04:34.048
But then, so what you emphasize actually for also during your talk,

00:04:34.128 --> 00:04:38.928
what is a four key perspectives, let's say, that you follow, right?

00:04:38.948 --> 00:04:44.568
One is this whole idea of how do you allocate neural structure or neural modules,

00:04:44.568 --> 00:04:49.388
We can talk about how to define that, but how do you allocate those strategically

00:04:49.388 --> 00:04:53.168
in the face of these challenges for survival, right?

00:04:53.288 --> 00:04:57.788
And then the second one, of course, how do you exactly control the parameters of your neurogenesis?

00:04:59.028 --> 00:05:04.828
The fourth one was how do you get coordination to actually build a multi-component system?

00:05:05.108 --> 00:05:11.728
And lastly, this whole role of embodiment and motivation as additional constraints on that process.

00:05:11.728 --> 00:05:17.208
So these are four big categories, if you want, of questions when we look at

00:05:17.208 --> 00:05:20.528
this whole issue of the development of a species-specific brain.

00:05:21.168 --> 00:05:26.388
But now if you would have to rank order those in terms of which of these four

00:05:26.388 --> 00:05:31.448
categories would be understood best and which of these four is actually of the

00:05:31.448 --> 00:05:33.688
greatest importance for understanding,

00:05:34.208 --> 00:05:37.788
how would you rank order the four? Yeah.

00:05:46.230 --> 00:05:52.830
Uh, boy, uh, let's see. I hope we get to edit some of this out here, but, um, yes.

00:05:53.910 --> 00:06:03.530
Uh, those four things I, I talked about were more, um, how the sort of empirical

00:06:03.530 --> 00:06:06.870
gathering of data that I did, um,

00:06:07.090 --> 00:06:09.810
you know, separate itself into categories.

00:06:09.810 --> 00:06:13.010
The real categories, I think, span those.

00:06:13.350 --> 00:06:20.350
So there's the one category which Jerison called, I think, originally proper

00:06:20.350 --> 00:06:26.650
mass or something, which is allocation of neural resources to what the animal's

00:06:26.650 --> 00:06:27.870
actually going to encounter.

00:06:27.870 --> 00:06:34.210
And that is both point number one and point number four.

00:06:34.210 --> 00:06:38.170
Okay so um so

00:06:38.170 --> 00:06:41.910
we find what looked like uh something of

00:06:41.910 --> 00:06:47.670
easily modifiable or presets or something like that in an evolutionary sense

00:06:47.670 --> 00:06:58.290
of am i going to be the kind of animal that that maximizes uh chemosensation um or um.

00:06:59.517 --> 00:07:05.677
Visual auditory or something like that. And it seems to be that over and over

00:07:05.677 --> 00:07:12.457
again, most likely independently from the very first bony fish and sharks,

00:07:12.697 --> 00:07:15.957
amphibians, reptiles, mammals,

00:07:16.117 --> 00:07:20.537
that animals keep diverging

00:07:20.537 --> 00:07:23.837
on those same differences

00:07:23.837 --> 00:07:27.217
again and again as a gross allocation

00:07:27.217 --> 00:07:30.297
of one of neuromass to

00:07:30.297 --> 00:07:34.057
one or the other in terms of processing resources okay um

00:07:34.057 --> 00:07:36.977
but then the last bit of of

00:07:36.977 --> 00:07:45.497
the talk which is that um most i'd say probably most of that given that bias

00:07:45.497 --> 00:07:48.457
and where you're going to get your sensory information comes is going to be

00:07:48.457 --> 00:07:54.237
coming from how the environment and motivation instructs the animal so So in

00:07:54.237 --> 00:07:56.357
terms of nervous system content,

00:07:56.777 --> 00:07:58.457
that's the big thing.

00:07:59.817 --> 00:08:09.917
Now, the other two are, I think, both about processing, and those are orthogonal.

00:08:09.917 --> 00:08:14.217
Orthogonal so um my my

00:08:14.217 --> 00:08:17.817
covert theory about what makes a

00:08:17.817 --> 00:08:21.777
vertebrate and why vertebrates took off at the 450 million

00:08:21.777 --> 00:08:28.077
500 years i did is they had in place sort of for the four big learning engines

00:08:28.077 --> 00:08:36.077
that were would be useful so this is uh which is things like um like cortex

00:08:36.077 --> 00:08:39.157
and hippocampus are both sort of association,

00:08:40.277 --> 00:08:48.797
information extractors and then we have reinforcement learning as embodied in basal ganglia.

00:08:50.057 --> 00:08:56.797
And the last sort of is the cerebellar kind of learning which is the,

00:08:57.897 --> 00:09:00.857
comparison and subtraction and optimization on that

00:09:00.857 --> 00:09:03.557
that dimension you find all those in the

00:09:03.557 --> 00:09:08.017
very first vertebrates um all together that's a bit of a surprise actually even

00:09:08.017 --> 00:09:13.897
to many professional neuroscientists who will say oh cortex that's just in mammals

00:09:13.897 --> 00:09:18.397
well i don't mean the cortex per se i mean the thing that the thing that the

00:09:18.397 --> 00:09:21.357
thing that cortex is doing is in all these other species.

00:09:21.817 --> 00:09:29.277
But your talk emphasized the conservation of these structures over time,

00:09:29.457 --> 00:09:31.577
but also the adaptability,

00:09:31.837 --> 00:09:37.637
and the adaptability was particularly around the relative sizes of parts of

00:09:37.637 --> 00:09:38.737
the brain, but even there,

00:09:39.017 --> 00:09:44.577
you know, the single greatest predictor of how big any bit of the brain is gonna

00:09:44.577 --> 00:09:47.137
be is how big the overall brain is, that's right?

00:09:47.137 --> 00:09:55.697
Yeah, so each part of the brain has its own rate of change with respect to total brain size.

00:09:56.117 --> 00:10:05.417
And apparently, no vertebrate has ever decided that the way to enlarge your brain is to,

00:10:05.417 --> 00:10:11.457
you know, double down on the number of motor neurons you have and nothing else.

00:10:11.457 --> 00:10:21.097
So what animals do, if their energetic situation or whatever allows them to support more brain,

00:10:21.317 --> 00:10:27.557
where that extra brain is going to go is in these sort of laterally placed association

00:10:27.557 --> 00:10:31.877
areas that look across sensory and motor systems.

00:10:31.877 --> 00:10:37.777
So there's an interesting question here around whether no vertebrates explored

00:10:37.777 --> 00:10:44.197
outside that space because it's not possible to go outside that space or it

00:10:44.197 --> 00:10:45.157
wouldn't be evolutionary.

00:10:45.777 --> 00:10:50.177
Advantageous, or is it just that there's certain aspects of this brain architecture

00:10:50.177 --> 00:10:53.137
which are locked in place which cannot now be changed easily?

00:10:53.517 --> 00:10:55.797
Yeah, that's a very interesting question.

00:10:58.122 --> 00:11:07.602
Paul Katz, who is at Georgia State University and past president of the Society

00:11:07.602 --> 00:11:10.982
for Neuroethology, had some data recently that just blew me away,

00:11:11.122 --> 00:11:14.482
which was sort of going to this particular question.

00:11:14.482 --> 00:11:17.862
So he studies marine mollusks, and of which there,

00:11:18.002 --> 00:11:22.182
I'm going to do violence to the numbers on these, but let's,

00:11:22.182 --> 00:11:30.602
so let's say there are 60,000 of them, give or take an order of magnitude, okay? Okay.

00:11:31.142 --> 00:11:39.262
And of those marine mollusks, only a very small fraction, some number,

00:11:39.382 --> 00:11:45.882
say a couple of orders of magnitude down, say 40 to 100, actually detached and have become mobile.

00:11:47.982 --> 00:11:51.442
And given the kind of work I've been doing with vertebrates,

00:11:51.522 --> 00:11:59.822
I fully expected to see that we would see sort of bilateral symmetry and swimming

00:11:59.822 --> 00:12:03.202
along by alternating contractions and.

00:12:05.082 --> 00:12:09.822
That's kind of the vertebrate optimal I had come to expect from seeing all those

00:12:09.822 --> 00:12:10.582
different kinds of things.

00:12:10.842 --> 00:12:13.782
What you do see couldn't be more different.

00:12:14.222 --> 00:12:19.742
So we have all these animals that have been attempting to swim for comparable

00:12:19.742 --> 00:12:23.722
periods of time and some of them are doing handsprings.

00:12:23.982 --> 00:12:27.222
Some of them are just looking like they're having some kind of seizure.

00:12:28.002 --> 00:12:31.982
Right. Some of them are doing by like, you know, and their job is,

00:12:31.982 --> 00:12:34.742
it's just like the evolutionary learning algorithms where they,

00:12:34.742 --> 00:12:39.062
you know, make animals in the

00:12:39.062 --> 00:12:43.522
computer reproduce if they just succeed in getting over some finish line.

00:12:44.313 --> 00:12:45.593
These are the same kind of things.

00:12:45.633 --> 00:12:50.533
And you see this same kind of wild differences in kinds of movement.

00:12:50.553 --> 00:12:55.193
And then I realized on seeing that stuff that truly not everything is evolvable.

00:12:56.013 --> 00:13:02.613
That these are animals that have become mobile and have, but...

00:13:03.833 --> 00:13:07.333
That's not a good starting point. Yeah, they've remained there as sort of singular

00:13:07.333 --> 00:13:09.993
and kind of comical examples, a lot of them.

00:13:10.913 --> 00:13:17.193
And so there is some data that would bear on this kind of question about what

00:13:17.193 --> 00:13:21.693
kinds of things permit changing.

00:13:21.693 --> 00:13:26.613
I guess if you read the classical account of, and by that I mean there's a book,

00:13:26.673 --> 00:13:31.373
Romer, which everyone reads when they study vertebrates and what they attribute

00:13:31.373 --> 00:13:35.653
that the sudden change in the number of extant species from,

00:13:35.793 --> 00:13:43.913
you know, 30 species to 40,000 species of lamprey-like animals to bony fish.

00:13:43.913 --> 00:13:50.673
They attribute that to the jaw, that you can now exploit so many different kinds

00:13:50.673 --> 00:13:52.633
of prey and eat so many different things.

00:13:52.793 --> 00:13:59.253
Well, jaw's all good, probably true, but there's a brain there too that suddenly appears.

00:14:00.533 --> 00:14:07.213
But the brain isn't so different maybe between ancestral jawless fish and the successful fish?

00:14:08.773 --> 00:14:14.793
Well, there's not so much forebrain. The whole parasympathetic-sympathetic division

00:14:14.793 --> 00:14:17.973
of the nervous system doesn't really come in until you get the jawed fishes.

00:14:18.153 --> 00:14:24.913
So there's been a lot of… But we don't really have any extant jawless fish of any sophistication.

00:14:24.933 --> 00:14:27.293
We have scavengers and bottom-feeding.

00:14:30.113 --> 00:14:31.153
True, but….

00:14:34.672 --> 00:14:38.892
You know, it certainly could be the case that there could be combinations that

00:14:38.892 --> 00:14:41.032
we haven't seen, but what we have there,

00:14:41.212 --> 00:14:46.152
the ones that did succeed, that have populated all these niches,

00:14:46.172 --> 00:14:51.852
do have extra brain features as well as just the jaws.

00:14:51.852 --> 00:14:56.892
And so I'm not saying that I know that to be the case, but there's sort of two

00:14:56.892 --> 00:14:58.512
points I'm trying to make here.

00:14:58.512 --> 00:15:09.512
One is that once you see Paul Katz's catalog of marine mollusks,

00:15:09.572 --> 00:15:15.932
you can see that without any particular proof at this point,

00:15:15.952 --> 00:15:20.212
that some things seem to be more potentially adaptable than others.

00:15:20.332 --> 00:15:23.692
And that not everything is possible from every starting point.

00:15:23.692 --> 00:15:27.052
And the other is for the traditional

00:15:27.052 --> 00:15:30.612
vertebrate account i think we should consider the brain and

00:15:30.612 --> 00:15:33.292
it's capable of not not just stop well it's the

00:15:33.292 --> 00:15:39.092
jaw so but but what's remarkable about the story of the brain is that this really

00:15:39.092 --> 00:15:44.732
quite complex structure was there so early on and it has it has been able to

00:15:44.732 --> 00:15:48.972
adapt while keeping that basic structure as animals for instance have moved

00:15:48.972 --> 00:15:52.452
out of the water and onto land, and then from there into the air,

00:15:52.632 --> 00:15:55.112
and then back into the water.

00:15:55.352 --> 00:16:00.352
Exactly. And it's done that without any change in the gross architecture.

00:16:00.632 --> 00:16:04.172
But there's lots of other changes around that architecture.

00:16:04.912 --> 00:16:10.292
But what I'm pointing out is that the neglected part of that original architecture

00:16:10.292 --> 00:16:20.152
may be something about this sort of multiplication or having several kinds of,

00:16:20.192 --> 00:16:21.272
let's call them learning engines,

00:16:21.332 --> 00:16:28.552
in that structure that may not really have been in a number of the other small organisms at the time,

00:16:29.092 --> 00:16:31.792
that maybe that's where we might be looking.

00:16:31.912 --> 00:16:35.392
So the preserved structure in all of those animals, for example,

00:16:35.392 --> 00:16:39.232
For example, if we take the four brain divisions,

00:16:39.692 --> 00:16:49.512
the thing called the medial pallium is recognizable as the hippocampus in mammals

00:16:49.512 --> 00:16:55.192
and is involved in rapid memory and navigation, those sorts of things.

00:16:55.192 --> 00:16:59.132
And to the extent that there is any large comparative basis,

00:16:59.252 --> 00:17:04.472
it's the same kind of general function in birds in the several studies.

00:17:04.532 --> 00:17:09.192
Also in fish, I'm not so sure about what the reptile base is.

00:17:09.252 --> 00:17:12.692
But here's one part of the forebrain.

00:17:14.219 --> 00:17:17.039
Um is remarkably consistent in sort of where

00:17:17.039 --> 00:17:19.859
it is and what it does over all that range so i think

00:17:19.859 --> 00:17:22.739
we can really look at at that kind of

00:17:22.739 --> 00:17:27.399
um comparative base in the in the forebrain now that we really couldn't before

00:17:27.399 --> 00:17:33.999
but you present a very quantitative approach to um towards trying to understand

00:17:33.999 --> 00:17:41.019
these invariant aspects of brain evolution and you're pointing us to our to your

00:17:41.079 --> 00:17:43.639
translatingtime.net domain,

00:17:44.019 --> 00:17:47.839
where you present data on 18 different species.

00:17:48.179 --> 00:17:55.119
And then you use different approaches like regression, you looked at sizes of

00:17:55.119 --> 00:17:59.359
different brain structures to try to get a handle on, okay, how invariant are

00:17:59.359 --> 00:18:01.919
these structures and their relations across these 18 species?

00:18:02.359 --> 00:18:06.959
So what is standing out in that relationship most in your opinion?

00:18:06.959 --> 00:18:13.739
The thing that utterly surprised us right from the beginning and still does,

00:18:13.939 --> 00:18:20.139
as we still keep making mistakes about hypothesizing the opposite,

00:18:20.239 --> 00:18:28.739
which is the absolute stability of mammalian brain development.

00:18:30.319 --> 00:18:36.259
So essentially what this model says

00:18:36.259 --> 00:18:43.679
and can do is that I can transform the developmental schedule of a mouse,

00:18:43.859 --> 00:18:47.599
and I'm talking about just the brain here from the time the first neurons are

00:18:47.599 --> 00:18:52.319
generated to sort of the start of the very first behavior.

00:18:53.499 --> 00:19:04.359
And I can simply turn a non-linear dial and with 99% accuracy predict when that's

00:19:04.359 --> 00:19:07.319
going to happen in a cat, in a monkey, in a human.

00:19:07.779 --> 00:19:12.579
And the detail of this translation is very deep.

00:19:12.759 --> 00:19:17.199
So when I'm saying translate the schedule, I mean, I can tell you when the Purkinje

00:19:17.199 --> 00:19:19.059
cells in the cerebellum are born.

00:19:19.099 --> 00:19:23.459
I can tell you when the cells in layer four of the metasensory cortex are born.

00:19:23.579 --> 00:19:27.539
I can tell you all these very very specific things about brain development across

00:19:27.539 --> 00:19:31.799
all these species with that amount of accuracy. Now this...

00:19:32.836 --> 00:19:38.356
This was originally surprising, and then I thought that changing sort of telescoping

00:19:38.356 --> 00:19:44.296
and compressing or shifting time things would be one of the major ways by which

00:19:44.296 --> 00:19:47.276
species differences would occur.

00:19:48.436 --> 00:19:52.036
And we have got a couple cases of that that I described,

00:19:52.356 --> 00:19:57.536
which is in the case of how much cortex you want versus olfactory bulb,

00:19:57.636 --> 00:20:00.496
and the other was how you change a retina to be nocturnal or diurnal,

00:20:00.576 --> 00:20:04.676
where you do get shifts of schedules with respect to each other.

00:20:04.956 --> 00:20:11.616
But overall, that mammalian brain development schedule stays rock steady. You don't change it.

00:20:12.076 --> 00:20:17.136
This is not my interpretation of these data, that is the data.

00:20:17.536 --> 00:20:21.476
It's nonlinear across species or within a species?

00:20:21.476 --> 00:20:26.036
So when I say turn a nonlinear dial, that means in order to transform the very

00:20:26.036 --> 00:20:31.696
early events of a mouse schedule into a monkey schedule, I don't have to change

00:20:31.696 --> 00:20:35.996
it very much, but to change, we get onto an exponential curve.

00:20:36.156 --> 00:20:42.536
But for the late events like takes first step, There's going to be much more

00:20:42.536 --> 00:20:49.156
relative duration between the points in the monkey scale or the human scale than the mouse scale.

00:20:49.256 --> 00:20:53.256
So there's not so much difference in the first events. There's big temporal

00:20:53.256 --> 00:20:58.656
differences in the late events, but this is just an exponential,

00:20:58.836 --> 00:21:01.456
predictable curve, if that makes sense.

00:21:01.456 --> 00:21:04.676
Right. So there's just a couple of parameters that you can capture any creature.

00:21:04.896 --> 00:21:11.256
Yeah. Yeah. But now, do you see this model as providing you the scaffold of

00:21:11.256 --> 00:21:14.756
a potential brain, or it really defines a brain?

00:21:15.356 --> 00:21:21.136
So I was supposed to be describing it as a sort of scaffold of a brain,

00:21:21.336 --> 00:21:33.376
that this maybe is a stable economical structure that is best...

00:21:34.835 --> 00:21:41.115
Uh you know maybe the one that positions itself best to make maximum use of experience,

00:21:41.815 --> 00:21:45.195
uh i think it's that um but

00:21:45.195 --> 00:21:48.115
that's what that's a positive way of looking at it another one might be to say

00:21:48.115 --> 00:21:54.115
that this is a fixed timetable you can't mess with it without i i used to think

00:21:54.115 --> 00:21:59.955
that i mean i that's a stephen j gold kind of argument which is that and i i

00:21:59.955 --> 00:22:02.895
have that title as a contrast in a lot of my papers,

00:22:02.995 --> 00:22:07.515
which is things like, is it developmental constraint or developmental structure? Right.

00:22:07.935 --> 00:22:15.235
And I keep coming back to 450 million years of defending the same structure. Right. Okay?

00:22:15.635 --> 00:22:18.755
I mean, I think we could have gotten out of it in that amount of time if it

00:22:18.755 --> 00:22:21.155
was a true constraint. So we can't do better.

00:22:21.495 --> 00:22:28.595
Yeah, there's a fitness peak in some way. Yeah, so I'm thinking of this more,

00:22:28.695 --> 00:22:33.495
I'm trying to push myself into thinking, Okay, how do I describe this as an optimal state?

00:22:33.615 --> 00:22:39.455
What is it optimizing as opposed to what is it, you know, why are we stuck with it?

00:22:39.875 --> 00:22:45.575
A lot of these basic body plan things are so actively defended on a genetic

00:22:45.575 --> 00:22:52.895
level by animals that it looks like not only are they not constraints of the

00:22:52.895 --> 00:22:55.355
sense of being stuck with a thing,

00:22:55.355 --> 00:23:03.375
that they're actively kept in place in the genetics of the animal.

00:23:03.495 --> 00:23:12.155
So that makes it reasonable and also falsifiable, too, that these are in some way optimal.

00:23:12.335 --> 00:23:20.655
And that's a lot of the change in the evo-devo approach to whole body and nervous

00:23:20.655 --> 00:23:25.355
system stuff is, okay, let's consider that this might be optimal. what is it optimizing?

00:23:25.715 --> 00:23:32.315
Right. But this observation is based on a number of descriptors that you use

00:23:32.315 --> 00:23:33.955
to look at brain development.

00:23:34.315 --> 00:23:38.055
It's all pretty much straightforward gross anatomy. Sure, exactly.

00:23:38.475 --> 00:23:43.715
But then you could make the argument actually it blinds you for other kinds

00:23:43.715 --> 00:23:47.755
of influences that might be more dependent on environment that might be more

00:23:47.755 --> 00:23:49.715
flexible or more dynamically regulated.

00:23:50.335 --> 00:23:51.835
Or you don't expect.

00:23:53.506 --> 00:23:56.646
I really am not making that distinction in any way.

00:23:56.746 --> 00:24:03.686
So I'm thinking that this is a way that lets you be flexible and dynamically integrated.

00:24:04.306 --> 00:24:08.146
Except that you're always on this fake time schedule. Yeah. Right?

00:24:08.206 --> 00:24:11.106
There's no way of escaping that one. Yeah.

00:24:12.426 --> 00:24:18.606
Because what I'm sensing is a potential conflict with the more general Evo Devo

00:24:18.606 --> 00:24:20.946
perspective that you also have.

00:24:21.126 --> 00:24:25.426
Yeah. But now we come out with a perspective that says, well,

00:24:25.486 --> 00:24:27.466
actually, this whole developmental program is just fixed.

00:24:27.526 --> 00:24:30.406
There's nothing you can do about it. Environment won't influence that.

00:24:31.006 --> 00:24:35.846
On the other hand, Barry Culp was telling us yesterday about the roles of stress in development.

00:24:36.586 --> 00:24:42.106
This might be a factor, but maybe you don't see that because the level of description

00:24:42.106 --> 00:24:45.166
doesn't allow you to extract these features.

00:24:47.586 --> 00:24:55.926
I think the talk you presented actually did balance some of that fixed constraint with some flexibility,

00:24:56.206 --> 00:25:02.546
because you talked then a lot about how delaying timing or delaying the onset

00:25:02.546 --> 00:25:05.646
of certain things, I guess, but that's within a window.

00:25:05.706 --> 00:25:09.566
You can't delay things indefinitely, but you can shift things around enough

00:25:09.566 --> 00:25:11.186
that that can have quite big changes.

00:25:13.689 --> 00:25:17.809
I can, for example, add two things. So

00:25:17.809 --> 00:25:25.889
one thing I brought up with the olfactory versus visual stuff is a sort of preset

00:25:25.889 --> 00:25:34.029
for a commonly encountered change in niche in animals that they would have sort

00:25:34.029 --> 00:25:36.329
of been filtered to have mechanisms to respond to.

00:25:36.329 --> 00:25:39.789
Um stress you can

00:25:39.789 --> 00:25:44.009
recharacterize this is is we're used to seeing a very stable environment we're

00:25:44.009 --> 00:25:48.269
used to seeing a very unstable environment and so it's interesting to me also

00:25:48.269 --> 00:25:56.149
that you see this a sort of suite of stress responses as as part of the uh this

00:25:56.149 --> 00:25:59.609
is this is below innate Okay,

00:25:59.969 --> 00:26:06.509
you know, so this is more in the, you know, the structure of the genome over

00:26:06.509 --> 00:26:12.749
evolutionary time that has the ability to respond to both kinds of environments.

00:26:15.329 --> 00:26:21.129
There's another thing I know, I tried to get started on, but haven't gotten

00:26:21.129 --> 00:26:26.989
too far, which is understanding critical periods and plasticity.

00:26:26.989 --> 00:26:30.949
So, I mean, this is in fact one of our current projects.

00:26:31.029 --> 00:26:38.349
So, I know nothing at this point about what the constraints are on these sorts of things.

00:26:38.489 --> 00:26:46.769
So, you know, is there really any, say,

00:26:47.029 --> 00:26:53.369
across mammal optimal period to learn certain kinds of knowledge that are good

00:26:53.369 --> 00:26:54.529
for particular structures?

00:26:55.269 --> 00:26:59.009
No one's really looked at it, tried to gather information in that way,

00:26:59.089 --> 00:27:01.269
or do you not do that kind of thing?

00:27:01.369 --> 00:27:08.049
For example, in birdsong, which has been something that's been studied a great

00:27:08.049 --> 00:27:09.169
deal for critical periods.

00:27:10.169 --> 00:27:12.589
So, um, uh.

00:27:14.139 --> 00:27:18.379
So a story for bird song that's often told is that the,

00:27:18.479 --> 00:27:25.619
you know, say the birds in this hemisphere or whatever come north and establish

00:27:25.619 --> 00:27:30.699
their nest and the eggs are laid and the nestlings are in the nest and they

00:27:30.699 --> 00:27:33.299
hear a song in the spring and they get,

00:27:33.339 --> 00:27:38.099
they sort of match up to their template and they learn the song and there's

00:27:38.099 --> 00:27:41.319
a critical period and the NMDA receptors come on and they come off.

00:27:41.319 --> 00:27:44.359
Okay, so maybe there's this critical period.

00:27:44.439 --> 00:27:49.999
But then it turns out that there's an unfortunate set of nestlings who are born

00:27:49.999 --> 00:27:53.339
in August, and they don't hear any song at all.

00:27:53.579 --> 00:27:59.719
And so what do you do? Do you just waste all that reproductive effort and grew a whole bunch of song?

00:27:59.899 --> 00:28:02.399
Well, it turns out it isn't like that at all.

00:28:03.199 --> 00:28:09.239
These animals put their critical period on hold until the next spring when they

00:28:09.239 --> 00:28:13.999
actually will hear some song. So this is something where the appropriate kind

00:28:13.999 --> 00:28:17.619
of input appears to initiate the critical period.

00:28:18.799 --> 00:28:23.759
So maybe the thing that would be general would not be having a certain time,

00:28:23.779 --> 00:28:25.439
but a certain kind of initiation.

00:28:26.879 --> 00:28:33.339
So self-initiated, self-terminated critical periods as opposed to things that are set in there. Right.

00:28:33.399 --> 00:28:39.919
So that would then allow environmental cues to actually trigger a part of a

00:28:39.919 --> 00:28:41.219
developmental program. Yeah.

00:28:41.939 --> 00:28:47.139
And you really have to understand that I'm starting from the position where

00:28:47.139 --> 00:28:53.759
I thought that everything was in play and that everything should have been as changeable as that.

00:28:53.879 --> 00:28:56.659
And there seems to be a whole lot of structural stuff that just isn't.

00:28:56.699 --> 00:28:59.599
And then we'll see what happens.

00:28:59.599 --> 00:29:05.939
But now in your description of evolution of isocortex versus olfactory bulb,

00:29:06.079 --> 00:29:12.179
right, over these 18 mammalian species that you looked at, you saw there's a very specific pattern.

00:29:12.519 --> 00:29:17.339
At first, all species are closely clustered. It's not that there is some variability

00:29:17.339 --> 00:29:19.879
within each species clustered.

00:29:20.099 --> 00:29:26.519
But you also interpret that in terms of some sort of continuum in this relationship

00:29:26.519 --> 00:29:29.719
between isocortex size, olfactory bulb size.

00:29:29.739 --> 00:29:33.639
As if it's some sort of trade-off, like more olfactory bulb, less isocortex.

00:29:33.939 --> 00:29:37.239
Is that really how you see it? A trade-off between these structures?

00:29:37.239 --> 00:29:39.859
Well, I showed you two kinds of data.

00:29:39.939 --> 00:29:45.999
So I showed you data on a whole lot of different mammals that showed a continuum

00:29:45.999 --> 00:29:52.419
both across and within taxonomic groups.

00:29:52.999 --> 00:29:57.059
So that would mean that, I'm going to say, looking...

00:29:59.934 --> 00:30:08.014
Looking within monkeys, we can find some monkeys that have virtually no olfactory

00:30:08.014 --> 00:30:12.194
bulb and limbic system at all, hardly, and some that have a fair amount.

00:30:13.254 --> 00:30:16.734
And we can find that pretty much in any one of our groups.

00:30:16.734 --> 00:30:23.834
Okay, so then we go back to, instead of 180 animals or so, we go back to our

00:30:23.834 --> 00:30:29.434
18 that we have the really elaborate developmental data on.

00:30:29.674 --> 00:30:34.454
And so we can find the examples of the animals that are low and high on olfactory

00:30:34.454 --> 00:30:35.634
versus cortex dimensions.

00:30:35.634 --> 00:30:41.294
Mentions and we say okay is there a timing component to that and i can't really talk much.

00:30:42.154 --> 00:30:46.234
Uh about um how much

00:30:46.234 --> 00:30:50.134
the they are or how continuous those animals

00:30:50.134 --> 00:30:53.814
are i suppose i could but i don't think it's really enough data and

00:30:53.814 --> 00:30:57.414
and so then i can find if i just ask what let

00:30:57.414 --> 00:31:00.434
me you know make a split of these into high and low

00:31:00.434 --> 00:31:03.454
olfactory versus cortex groups um is

00:31:03.454 --> 00:31:07.394
there a difference in how they develop and yes there is so

00:31:07.394 --> 00:31:10.414
the the primates and the carnivores uh

00:31:10.414 --> 00:31:14.154
with the high cortex delay producing

00:31:14.154 --> 00:31:20.634
their cortex until later and that makes more of it because they have more time

00:31:20.634 --> 00:31:24.714
to develop their precursors so with the sticking point for me here is that is

00:31:24.714 --> 00:31:29.134
there any kind of intrinsic constraint in this developing brain that's okay

00:31:29.134 --> 00:31:31.894
if i'm allocating more resources to one structure,

00:31:32.114 --> 00:31:36.294
then let's say there's a metabolic cost and therefore I cannot grow another

00:31:36.294 --> 00:31:38.214
structure equally well.

00:31:38.334 --> 00:31:43.234
So that there's always a sort of, from a pure morphogenesis perspective, there are constraints.

00:31:44.094 --> 00:31:47.574
On the other hand, you go and say, no, every structure develops as an independent

00:31:47.574 --> 00:31:51.394
module triggered by environmental conditions, the niche you're in.

00:31:51.494 --> 00:31:55.394
So in principle, I could grow a huge olfactory bulb and a big cortex.

00:31:55.394 --> 00:32:01.334
If my environment, and actually, carnivores, some carnivores will do that, right?

00:32:01.474 --> 00:32:04.694
So where are we in those two interpretations?

00:32:06.254 --> 00:32:12.794
So brain is really expensive. So I've actually written a little bit about this.

00:32:12.854 --> 00:32:15.114
You can contrast two kinds of explanations.

00:32:16.234 --> 00:32:24.834
So one is if brain is expensive, then it's reasonable to get the kind of negative

00:32:24.834 --> 00:32:27.094
correlation that we see there.

00:32:27.094 --> 00:32:33.394
So if you have a high cortex value, you're relatively more likely to have a

00:32:33.394 --> 00:32:35.814
low limbic olfactory one.

00:32:38.614 --> 00:32:43.134
So when you have like an energetic or caloric restraint or something like that

00:32:43.134 --> 00:32:47.034
would be something you'd find. Another is a mechanistic constraint,

00:32:47.374 --> 00:32:51.714
which is something I've been looking at.

00:32:51.794 --> 00:32:56.754
I showed a picture that showed the fact that the thing that gives rise to the

00:32:56.754 --> 00:33:00.834
olfactory cortex and the cortex and the hippocampus are,

00:33:00.994 --> 00:33:10.174
so the neocortex is sitting right between the hippocampus and the olfactory

00:33:10.174 --> 00:33:11.714
cortex embryologically.

00:33:11.714 --> 00:33:17.194
And it looks like it would be just so easy, let's take that primordial tissue

00:33:17.194 --> 00:33:18.494
and give it more to the cortex.

00:33:18.974 --> 00:33:23.714
Let's take that and give it more to the other two, which are immediately adjacent to it.

00:33:24.794 --> 00:33:27.334
That implies a zero-sum game.

00:33:28.454 --> 00:33:35.494
That if it's that mechanism and you do it that way, it should always be push-pull like that.

00:33:35.654 --> 00:33:41.374
Now, the evidence I can offer against that is that it's only mammals animals?

00:33:42.191 --> 00:33:47.331
That have the negative correlation that I know of so far. So is that particular

00:33:47.331 --> 00:33:51.051
example, but are you then looking a bit too late in embryology?

00:33:51.411 --> 00:33:54.951
There's nothing earlier in embryology than that. This is really early,

00:33:54.971 --> 00:33:55.711
isn't it? Yeah, that's right.

00:33:55.891 --> 00:34:00.191
Because you were talking about the migration of the precursor cells,

00:34:00.871 --> 00:34:02.631
and this is at that stage.

00:34:02.771 --> 00:34:06.251
Yeah, so there is no earlier in which we could actually identify something that's

00:34:06.251 --> 00:34:07.431
going to give rise to the cortex.

00:34:07.431 --> 00:34:12.571
Okay, but one of the really interesting things I think you were showing was

00:34:12.571 --> 00:34:19.091
the constraint that the developmental process has on the potential for evolution,

00:34:19.371 --> 00:34:23.951
because you were saying that it was only the lateral parts of this embryological

00:34:23.951 --> 00:34:30.371
structure that had the potential to really change, and the more central parts were fairly fixed.

00:34:30.511 --> 00:34:36.391
Is that right? Yeah, so this is just a description of the data where if you

00:34:36.391 --> 00:34:40.111
lay out the embryonic brain on a front to back,

00:34:40.291 --> 00:34:50.231
middle to edge, that it turns out that how long the cells divide during early

00:34:50.231 --> 00:34:52.071
embryogenesis depends on position.

00:34:53.131 --> 00:34:57.551
So the closer you are to the edge, the longer, the closer you are to the front,

00:34:57.671 --> 00:35:02.191
the longer. Right. So this is just describing what's going on.

00:35:02.191 --> 00:35:05.151
But that description must be capturing some constraint, presumably,

00:35:05.331 --> 00:35:07.211
that we don't perhaps understand very well.

00:35:07.971 --> 00:35:12.011
I don't know if it's a, you know, I don't think it's a property of embryonic

00:35:12.011 --> 00:35:15.891
tissues to divide a lot at the edge or something. I've never heard of anything like that.

00:35:16.171 --> 00:35:20.111
But you know, so maybe there is something like that.

00:35:20.111 --> 00:35:29.051
But what I would say is sort of an interesting overall interpretation of this

00:35:29.051 --> 00:35:31.291
is to come at it just opposite.

00:35:31.571 --> 00:35:38.091
Okay, so what you do is you set up an embryonic structure that gives you variation on some dimension.

00:35:38.491 --> 00:35:43.591
And in this case, it's variation in the numbers of cells in the array that that

00:35:43.591 --> 00:35:45.071
structure is going to produce.

00:35:45.071 --> 00:35:54.291
Then you allocate function to that location so if you for example um,

00:35:56.081 --> 00:36:00.741
One thing that's quite interesting about this, there are two places in human

00:36:00.741 --> 00:36:07.081
brains and mammalian brains that always continue to produce neurons throughout life. What are those?

00:36:07.941 --> 00:36:12.101
That's the hippocampus and the olfactory bulb.

00:36:12.281 --> 00:36:16.901
Where are those? Those are sitting on exactly that edge there.

00:36:17.061 --> 00:36:20.681
Right. So if you look at bird brains,

00:36:20.901 --> 00:36:27.041
they produce neurons in many more places throughout life, but all you move in

00:36:27.041 --> 00:36:32.201
is just a little bit more towards the midline and pick up the structures that

00:36:32.201 --> 00:36:33.401
are sitting on that edge.

00:36:33.461 --> 00:36:37.641
If you're a fish, you're essentially generating all the brain throughout life,

00:36:37.781 --> 00:36:42.461
but you are generating sort of relatively more of it on those lateral edges.

00:36:42.461 --> 00:36:51.241
And I think the possibility is that you then take that variation in the size that it's going to be,

00:36:51.261 --> 00:36:54.481
in the potential energetic cost you're going to put into it,

00:36:54.501 --> 00:37:01.161
and then you can put function into it by designating those cells there in some different way.

00:37:01.421 --> 00:37:05.561
Right. But these are precursor cells. They've yet to specialize into particular

00:37:05.561 --> 00:37:08.061
neuron types. They've yet even to migrate into position.

00:37:08.741 --> 00:37:14.061
Well, I mean, I showed the fate map of this thing is, as we're describing it

00:37:14.061 --> 00:37:16.001
now in verbose, very fixed.

00:37:16.421 --> 00:37:19.461
So the medial part is always going to be motor neurons.

00:37:19.661 --> 00:37:23.361
And the next step over is going to be the visceral motor neurons.

00:37:23.541 --> 00:37:30.141
And so those locations mean something very specific in terms of what neuron

00:37:30.141 --> 00:37:32.781
is going to be being generated also means a duration.

00:37:33.241 --> 00:37:35.881
But when this all started out, maybe...

00:37:37.384 --> 00:37:43.324
The fact that assignment of type in the course of development comes after the

00:37:43.324 --> 00:37:47.204
decision of how long you're going to be generated means that it could be you

00:37:47.204 --> 00:37:55.984
assigned sort of type after the sort of size of the thing was entered into the equation.

00:37:56.324 --> 00:38:00.804
But I mean, it comes back to this question of how gridlocked is the design of the brain.

00:38:00.804 --> 00:38:07.284
And what essentially we're saying is that there's certain things that are decided

00:38:07.284 --> 00:38:10.344
early on in development, maybe, and if you try to change anything there,

00:38:10.564 --> 00:38:14.164
it might have lots of knock-on consequences. Yeah, so there's some of those things.

00:38:14.224 --> 00:38:19.944
But look, the things that do get big are the very things that stay out of gridlock.

00:38:20.064 --> 00:38:25.344
So the ones that are on the lateral edge are pretty much uniformly multisensory,

00:38:25.444 --> 00:38:32.384
multimotor, you know, can control different effector systems and are the very

00:38:32.384 --> 00:38:35.704
parts of the brain, with the exception of the very specific olfactory cortex,

00:38:35.924 --> 00:38:38.644
that are the most plastic and changeable in their functions.

00:38:39.184 --> 00:38:46.924
So you allocate more space to the specifically multimodal things,

00:38:47.064 --> 00:38:49.064
which can be allocated to anything.

00:38:49.424 --> 00:38:53.204
Yeah. But there's an interesting conclusion to that maybe, because on the one

00:38:53.204 --> 00:38:57.404
hand, I think it's also important to take into account the morphological constraint

00:38:57.404 --> 00:38:59.364
imposed by having a skull that you have to fill.

00:39:00.324 --> 00:39:03.144
Well, the brain generates the skull, not the other way around.

00:39:03.724 --> 00:39:08.364
But these things develop together. Yeah, yeah. And there will be also mechanical

00:39:08.364 --> 00:39:10.464
constraints on the developing brain.

00:39:10.744 --> 00:39:14.544
And you better lay down your brainstem before you lay down your cortex.

00:39:14.744 --> 00:39:16.404
Otherwise, you cannot pack it in there anymore.

00:39:16.644 --> 00:39:21.344
So this already defines a certain logical order, you would think.

00:39:21.704 --> 00:39:28.504
But then as you move out later in development, so the more primitive structures

00:39:28.504 --> 00:39:32.984
are laid down, it's not a surprise you might end up with the most nonspecific structures.

00:39:33.264 --> 00:39:37.964
Because these also should be the hyperplastic structure. because they are more

00:39:37.964 --> 00:39:44.284
dependent on somatic time to wire themselves into that system because they're

00:39:44.284 --> 00:39:45.724
under constraint, if you want.

00:39:46.304 --> 00:39:50.744
There's less guidance that you can give them. So maybe this already then tells

00:39:50.744 --> 00:39:54.024
you why these more cortical-like structures,

00:39:54.324 --> 00:39:58.404
these hyperplastic multimodal associative structures, are then more lateral

00:39:58.404 --> 00:40:02.544
and at the outside of a developing brain because they're actually,

00:40:02.624 --> 00:40:04.124
these are easy to specify.

00:40:04.124 --> 00:40:08.144
And you then leave it to their developmental to their learning capability to

00:40:08.144 --> 00:40:11.144
wire themselves up with the rest of the system would that make sense to you?

00:40:11.284 --> 00:40:14.684
Yeah there's another way of thinking about that which is quite parallel which

00:40:14.684 --> 00:40:22.524
is what a bunch of computer scientists thinking about control systems have done and so.

00:40:23.512 --> 00:40:30.352
Um, you know, so there's several things, uh, you know, several groups that have

00:40:30.352 --> 00:40:34.652
come down on, um, this same kind of organizational principle.

00:40:34.752 --> 00:40:40.612
So, so if you want to make a device that can describe sort of catastrophic loss

00:40:40.612 --> 00:40:44.192
and sudden gains, okay, how do you build it?

00:40:46.192 --> 00:40:51.252
Well what what you want to do is to keep its basic functions like getting around

00:40:51.252 --> 00:40:57.832
and recognizing things kind of untouched and the last thing you want to do if

00:40:57.832 --> 00:40:58.712
you're going to make a big,

00:40:59.232 --> 00:41:03.172
fancy new brain is to have you know your motor neurons on the one hand they're

00:41:03.172 --> 00:41:06.952
going to do something and then you have your sensory neurons and then between

00:41:06.952 --> 00:41:11.392
those you interpose some gigantic processing thing.

00:41:11.592 --> 00:41:15.912
And what you've succeeded in doing is now slowing down this organism so much

00:41:15.912 --> 00:41:18.432
that it will never, ever survive anything whatsoever.

00:41:19.232 --> 00:41:23.712
So what people found a much better control architecture to be is to keep those

00:41:23.712 --> 00:41:28.172
kinds of essential motor organizational functions by themselves,

00:41:28.332 --> 00:41:32.112
and you build a brain beside that brain. Right, okay. Okay?

00:41:32.452 --> 00:41:36.432
Yeah. And that's what's going on, I think, a better description here of,

00:41:36.432 --> 00:41:42.672
So, you're modeling, you're building a model of your brain, sort of this predictive

00:41:42.672 --> 00:41:45.892
thing, and evolution says, oh, that seems like a good idea, too.

00:41:45.952 --> 00:41:47.772
I'm glad you came to your side and thought of that, you know,

00:41:47.772 --> 00:41:53.692
that you are now being able to plan and simulate and integrate things while

00:41:53.692 --> 00:41:55.632
still kind of carrying on as usual.

00:41:55.892 --> 00:41:59.512
So, that means you lay down these midline structures, and then you sort of pad

00:41:59.512 --> 00:42:02.992
it with a hyperplastic vortex-like structure.

00:42:03.152 --> 00:42:05.792
Yeah, and it's literally beside in two ways. Yeah, exactly.

00:42:06.572 --> 00:42:11.012
So it's not, you know,

00:42:11.012 --> 00:42:18.512
somehow I find it easier or more pleasant to think about this sort of building

00:42:18.512 --> 00:42:23.592
a brain beside the brain than just having extra stuff lying around. Sure.

00:42:23.952 --> 00:42:30.792
But perhaps it comes down to the same. so so what so looking across all these

00:42:30.792 --> 00:42:36.972
pieces you've analyzed um what would you now see as the blueprint of the mammalian brain,

00:42:39.152 --> 00:42:43.712
um so we have um uh.

00:42:45.972 --> 00:42:55.912
The whole spinal motor sensory core that um you know takes care of all sort

00:42:55.912 --> 00:43:00.532
of essential movement and eating and breathing.

00:43:00.832 --> 00:43:05.912
And then we kind of add on some limbs if we're going to be fancy that our,

00:43:08.119 --> 00:43:16.439
sort of the fundamental operating arrangement, then I'm very much fond of a

00:43:16.439 --> 00:43:21.779
guy named Bjorn Merker and his views of how to,

00:43:22.419 --> 00:43:26.119
but he's getting into consciousness, but we don't have to discuss that so much.

00:43:26.479 --> 00:43:31.239
We know Bjorn very well. We just spent two weeks with him in Woods Hole. Okay, so great.

00:43:31.659 --> 00:43:39.639
So he views the midbrain as the place where all this basic integration comes

00:43:39.639 --> 00:43:45.359
together to make a sort of a sketch of operations for the animal. So what's in front of me?

00:43:46.079 --> 00:43:48.419
What can I do with it? What do I want to do?

00:43:51.139 --> 00:43:58.359
Then sort of coming from the other direction, we have the whole visceral brain

00:43:58.359 --> 00:44:01.459
that knows about, okay, what is my state?

00:44:02.519 --> 00:44:07.299
What sort of future state would I like? Am I trying to mobilize energy now or save it?

00:44:07.419 --> 00:44:12.159
Or, you know, what do I want to show other individuals about what my energy state is?

00:44:12.319 --> 00:44:14.959
And then I'm going to combine that with that same sketch.

00:44:16.359 --> 00:44:20.419
This is something that's part of the vertebrate makeup and highly plastic.

00:44:20.539 --> 00:44:23.059
It's going to be quite different from one species to the next,

00:44:23.119 --> 00:44:27.219
how energy is going to be allocated and how fast and towards what.

00:44:27.219 --> 00:44:35.579
And then we have any of these gigantic learning loops sitting around this whole thing.

00:44:35.599 --> 00:44:43.119
One is the slow auto-associating cortex thing, the fast auto-associating hippocampus.

00:44:43.119 --> 00:44:45.939
The uh one who's going to take the output of both

00:44:45.939 --> 00:44:48.999
those things together um the reinforcement circuitry and

00:44:48.999 --> 00:44:52.759
says okay which which of these combinations actually helped me and which do

00:44:52.759 --> 00:44:58.039
i wish to repeat as an animal and then the cerebellar like circuits that uh

00:44:58.039 --> 00:45:01.019
um take plans and optimize them

00:45:01.019 --> 00:45:08.919
um so so basically i see this uh motor motivational core with uh these,

00:45:09.839 --> 00:45:16.359
these second brains sitting to the side and computing the state of that basic

00:45:16.359 --> 00:45:17.939
operating system, I guess.

00:45:18.099 --> 00:45:21.359
And that would be it. So four second brains.

00:45:21.779 --> 00:45:27.099
Yeah. So which animal of all the species that you, in your database,

00:45:27.179 --> 00:45:34.319
which animal then gives us the purest reflection of that blueprint, if you want?

00:45:34.619 --> 00:45:40.559
Everyone. but let's say some would have reduced some parts of it they might

00:45:40.559 --> 00:45:42.279
have exaggerated other parts mm-hmm.

00:45:45.352 --> 00:45:48.472
The house cat, I don't know. Okay.

00:45:49.472 --> 00:46:01.092
No, I mean, since at least the mammals all are on the same general trajectories,

00:46:01.932 --> 00:46:04.212
it's really almost impossible to answer that question.

00:46:04.812 --> 00:46:10.432
But one thing that is obviously talked about is the change of size in the cortex,

00:46:11.412 --> 00:46:14.972
perhaps more than changes in these other second brain systems.

00:46:15.572 --> 00:46:18.972
Do you think too much is made of that? But it's not the case that it's the,

00:46:19.112 --> 00:46:25.932
you know, so it's on its exactly expected allometric line, as is the cerebellum.

00:46:26.352 --> 00:46:31.232
A lot of people make a lot of, you know, whether we should count neurons or

00:46:31.232 --> 00:46:35.632
volume or something, or caloric expense, you basically need to count them all.

00:46:35.812 --> 00:46:40.272
You know, so how many neurons is one measure of how big a structure is.

00:46:40.792 --> 00:46:46.232
How actually big it is is another measure of how big it is. or how many synapses you have.

00:46:46.712 --> 00:46:56.412
But the Harry Jerison story was that so-called higher mammals had bigger brains than other mammals.

00:46:56.752 --> 00:47:01.932
So you can dissociate brain size from body size, right?

00:47:02.052 --> 00:47:07.452
But you cannot dissociate internal brain structure size from brain size.

00:47:07.672 --> 00:47:15.092
So a certain brain will always, if it's a primate, I mean, let's set the olfactory

00:47:15.092 --> 00:47:17.372
parameter, okay, at the outset.

00:47:18.372 --> 00:47:24.172
We'll have, we have exactly the size cortex we should have for our size of brain.

00:47:24.292 --> 00:47:31.112
If we were a dolphin and we have more, a bigger brain, we have more cortex than us, proportionately.

00:47:32.200 --> 00:47:36.740
So there's no special selection on the cortex. But then there's the strong claim

00:47:36.740 --> 00:47:39.360
that in hominid evolution… It's not true.

00:47:40.140 --> 00:47:42.300
Okay. It's the right size.

00:47:43.280 --> 00:47:48.540
I mean, in relative terms, there's an invariance. There was a change in brain size.

00:47:48.860 --> 00:47:51.880
So, yeah, the thing is that the brain size, we have a really,

00:47:51.920 --> 00:47:53.860
really big brain for our body size.

00:47:54.000 --> 00:47:57.920
We have exactly the cortex size that we should have for our brain size.

00:47:58.820 --> 00:48:04.600
Okay. But is that a constraint that if you need, just to take the corticocentric

00:48:04.600 --> 00:48:08.500
view, which I don't actually hold, but sort of devil's advocate,

00:48:08.880 --> 00:48:13.320
if I want a bigger cortex and I have these developmental constraints,

00:48:13.460 --> 00:48:16.140
I just have to build a bigger brain. There's no other way around it.

00:48:18.060 --> 00:48:21.480
That's what evolution says so far.

00:48:21.480 --> 00:48:26.480
So it could be read in that way, if I'm wedded to my view that humans have this

00:48:26.480 --> 00:48:32.180
fantastic neocortex, and that we just grew extra bits of brain that we maybe

00:48:32.180 --> 00:48:34.940
don't use so much in order to make that possible.

00:48:35.480 --> 00:48:39.020
No, but there's something that, there's an inconsistency now here,

00:48:39.140 --> 00:48:42.200
because if it's always relative to overall brain size.

00:48:42.340 --> 00:48:46.900
Yes, there's a real problem. Earlier, but earlier we discussed that you said,

00:48:46.980 --> 00:48:52.760
no, you can actually have relative differences between a limbic brain,

00:48:53.180 --> 00:48:55.960
the limbic cortex, and the isocortex.

00:48:56.420 --> 00:49:02.020
Yeah, so I prefaced this whole thing with let's set the limbic factor.

00:49:02.640 --> 00:49:07.840
Ah, okay. You were smart. Yes, good. Okay. The second component.

00:49:08.560 --> 00:49:11.280
Okay, fair enough. Okay, fair enough.

00:49:11.280 --> 00:49:17.900
So that second component says that, yeah, my cortex scales with my brain size,

00:49:18.080 --> 00:49:25.080
but also I can be a cortical-oriented species, or I can be an olfactory-lymbic-oriented species.

00:49:25.320 --> 00:49:27.800
And that accounts for how much of the variance, roughly?

00:49:29.060 --> 00:49:35.980
3% of it. Really? Oh, okay. 3% is a lot of variance, considering the range that we have here.

00:49:36.200 --> 00:49:39.240
That's very tiny. A lot of volume. Well, it's a small amount of variance,

00:49:39.320 --> 00:49:40.440
it's a large amount of tissue.

00:49:41.280 --> 00:49:45.120
If you're considering the difference between, you know.

00:49:46.622 --> 00:49:49.342
Uh mediums well i showed i showed

00:49:49.342 --> 00:49:53.522
a picture of uh of of a owl monkey's

00:49:53.522 --> 00:49:57.182
brain and a goodie brain of exactly the same mass and

00:49:57.182 --> 00:50:00.262
you know and and so the one is a cortex specializer and

00:50:00.262 --> 00:50:05.062
the cortex is hanging all over the side so you can't see the olfactory bulb

00:50:05.062 --> 00:50:08.402
and you can't see the cerebellum and all that in the owl monkey because the

00:50:08.402 --> 00:50:13.442
cortex has overgrown it but in this very same sized agouti you get a very good

00:50:13.442 --> 00:50:17.702
view of the olfactory bulbs and the cerebellum and everything,

00:50:17.802 --> 00:50:20.342
just because the cortex has it. Which is a rodent, right?

00:50:20.602 --> 00:50:22.802
Yeah, it's a big South American rodent.

00:50:23.622 --> 00:50:30.702
And so 3% sounds little, but if you look at those brains, that's a perfectly

00:50:30.702 --> 00:50:32.482
impressive difference.

00:50:32.902 --> 00:50:36.122
And I guess, staying with my devil's advocate position,

00:50:36.122 --> 00:50:43.342
some people would also argue that cortex has become specialized in other ways

00:50:43.342 --> 00:50:48.582
in primates for instance you know we have six layers of cortex like every other

00:50:48.582 --> 00:50:49.762
mammal but we seem to have.

00:50:51.242 --> 00:50:56.022
Richer networks within some of those cortical layers i mean do you do you buy

00:50:56.022 --> 00:51:01.042
into any of that i think you had a slide showing that the layer 2-3 was expanded

00:51:01.042 --> 00:51:05.722
but the point of my slide Registered network, interconnectivity at least.

00:51:05.962 --> 00:51:16.062
But I was trying to show that this is how the gradient of the cortex plays out over different brains.

00:51:16.142 --> 00:51:24.942
So yeah, in the set of animals I have in this set, the human has the largest

00:51:24.942 --> 00:51:27.962
cortex, but I don't have any dolphins or whales in there.

00:51:27.962 --> 00:51:34.662
So I can't really say that, that we somehow they're primates or anything,

00:51:34.722 --> 00:51:39.082
hold the prize for most complex network. And I'm not sure exactly what.

00:51:41.167 --> 00:51:45.307
Means, except sort of a self-reifying, my, we're complex, look at that thing, it's complex.

00:51:45.567 --> 00:51:49.427
Well, I think if Henry Kennedy was here, he'll be here next week,

00:51:49.487 --> 00:51:51.647
and he can contradict this.

00:51:52.207 --> 00:51:58.967
He would say that primates have this richer, within the layers,

00:51:59.027 --> 00:52:04.607
they have these circuits that take advantage of these additional cells that

00:52:04.607 --> 00:52:06.867
you have there. So you have a few sub-layers.

00:52:07.427 --> 00:52:12.187
I guess so the question is whether this is some virtue of being a primate or

00:52:12.187 --> 00:52:13.627
virtue of having a large cortex.

00:52:13.907 --> 00:52:17.027
Yeah. And that we just don't know yet.

00:52:17.187 --> 00:52:23.567
You also said that with respect to niche specificity, this might also relate to this point,

00:52:23.807 --> 00:52:27.927
that actually there are other processes at work as well that might be linked

00:52:27.927 --> 00:52:32.927
to then how that organism interacts with its niche, which might be control of hematosis,

00:52:33.387 --> 00:52:36.447
thalamic drive onto a cortical structure.

00:52:36.867 --> 00:52:42.387
Which might vary, and in general, activity-dependent volume change, right?

00:52:42.407 --> 00:52:44.987
So that might mean that from a developmental perspective, you have,

00:52:45.027 --> 00:52:51.087
let's say, a prototypical scaffold that then gets biased by how that niche and

00:52:51.087 --> 00:52:54.867
the embodiment is sort of driving the scaffold using these three principles.

00:52:55.207 --> 00:53:02.027
Would you buy that? Yeah, I mean, so taking this basic set of layers and then

00:53:02.027 --> 00:53:07.047
embedding it in different kinds of experience or different kinds of early instruction,

00:53:07.247 --> 00:53:09.187
I imagine you get all kinds of different things.

00:53:09.667 --> 00:53:13.167
It might then account for these differences. There might not be no contradiction.

00:53:13.727 --> 00:53:17.607
Yeah, I think so. And there's a lot of really basic stuff that we don't know.

00:53:17.647 --> 00:53:20.727
If you look at dolphin cortex, for example,

00:53:21.227 --> 00:53:34.167
it puts a little bit more total volume into area and less into the cortex depth and I mean what.

00:53:34.883 --> 00:53:38.443
My group and I, we've done a lot of modeling of this kind of thing,

00:53:38.483 --> 00:53:43.783
and you have only to change that sort of quit fraction in early development

00:53:43.783 --> 00:53:51.963
in the cortex the very slightest amount to sort of direct stuff into more neurons

00:53:51.963 --> 00:53:54.143
per cortical column versus more area.

00:53:54.143 --> 00:54:02.323
So I don't think we really know very much yet about just what the consequences

00:54:02.323 --> 00:54:05.463
of small changes in early developmental parameters are.

00:54:05.663 --> 00:54:10.143
And I've never understood why

00:54:10.143 --> 00:54:16.723
more layers in a cortex was supposed to be somehow intrinsically better.

00:54:18.863 --> 00:54:26.763
Um you know we still

00:54:26.763 --> 00:54:30.003
have the same um hippocampus that

00:54:30.003 --> 00:54:33.463
everybody else has and including the fish and and

00:54:33.463 --> 00:54:36.303
seems to do just fine i don't i mean i just

00:54:36.303 --> 00:54:39.263
don't see i mean if someone would come up and

00:54:39.263 --> 00:54:42.063
and show me and i defy you to find someone who

00:54:42.063 --> 00:54:45.043
has that here we have a five layered structure

00:54:45.043 --> 00:54:48.763
and look what it can do but I'm going to make six layers and oh man now we got

00:54:48.763 --> 00:54:53.663
calculus I mean I don't think so so it's really I mean it would be nice to actually

00:54:53.663 --> 00:54:58.023
see some demonstration but now it's all sort of I you know large numbers mean

00:54:58.023 --> 00:55:02.083
something more complex you did in your talk describe some,

00:55:03.518 --> 00:55:08.078
changes across cortex particularly talked about a gradient of increased compression

00:55:08.078 --> 00:55:12.418
going from the back of the brain to the front of the brain and then you talked

00:55:12.418 --> 00:55:18.098
about the front of the brain having more fan in more systems talking into the front of the brain,

00:55:18.658 --> 00:55:24.698
and which i got from henry kennedy's yeah study i wanted to put so and you can

00:55:24.698 --> 00:55:29.678
imagine that bigger brains are going to have more steps of compression exactly

00:55:29.678 --> 00:55:31.098
you don't have to imagine that they do.

00:55:31.278 --> 00:55:35.138
So by the time you get to frontal cortex, you've got more abstraction. Exactly.

00:55:35.558 --> 00:55:39.718
Yeah. That's how you get calculus. Yeah. Yeah. But you just conflated.

00:55:40.058 --> 00:55:41.758
I'm happy you explained that to us, Tony.

00:55:44.378 --> 00:55:49.878
We were talking about layers, you know, whether six layers of cortex is better

00:55:49.878 --> 00:55:55.538
than four in implementing calculus versus the number of steps that in a sort

00:55:55.538 --> 00:55:57.978
of hierarchy, which is a different thing altogether.

00:55:58.158 --> 00:56:04.018
So I can easily make an argument as to why embellishing a hierarchy might be

00:56:04.018 --> 00:56:08.318
a better thing for extracting more and more abstract explanation.

00:56:08.358 --> 00:56:09.818
And I wouldn't have to come up with it myself.

00:56:09.978 --> 00:56:14.998
I could come up with all kinds of computational models that people have made on exactly this point.

00:56:16.818 --> 00:56:23.458
So that's why finding this hierarchy from a large number of neurons in the back

00:56:23.458 --> 00:56:28.738
of the cortex to a very small number in the front and this progressively greater compression,

00:56:29.038 --> 00:56:35.358
the bigger the brain gets, sort of maps on to the computational work that people

00:56:35.358 --> 00:56:36.758
have done really nicely.

00:56:37.178 --> 00:56:40.558
You know, and so I'm not making fun of the layer thing so much,

00:56:40.618 --> 00:56:45.238
but there's no comparable literature that says, you know, five layers in a column

00:56:45.238 --> 00:56:48.318
allows me to do something differently.

00:56:49.431 --> 00:56:53.211
That I couldn't do with four. I've just never seen anything even take that on.

00:56:53.451 --> 00:56:55.251
Right. You know, so it could be.

00:56:55.891 --> 00:57:02.811
But in some sense, you also made the point that if I have sort of this midline

00:57:02.811 --> 00:57:06.411
controller, and then we have all these add-on learning machines,

00:57:07.371 --> 00:57:13.171
then you said, oh, if you then take Cortex, it's sort of equipotential, right?

00:57:13.631 --> 00:57:16.311
Starting sort of equipotential, I guess, in the smaller brains.

00:57:16.311 --> 00:57:19.751
But you also made that point where you said, look, what's special about,

00:57:19.911 --> 00:57:23.891
let's say, a language area, if you just look at it from, let's say,

00:57:23.931 --> 00:57:28.151
a morphological anatomical perspective, is there anything special about it?

00:57:28.211 --> 00:57:33.051
So you seem to be making this claim that cortex has this sort of,

00:57:33.051 --> 00:57:35.051
this infinite, not infinite,

00:57:35.271 --> 00:57:40.691
but this very hyperplastic properties that would allow it to sort of tune to

00:57:40.691 --> 00:57:43.351
any kind of information that it is exposed to.

00:57:43.351 --> 00:57:49.431
So, equipotentiality is really, for you, an important principle to understand

00:57:49.431 --> 00:57:50.631
how the system operates?

00:57:52.671 --> 00:57:56.091
Well, I'm going to sort of go empirical on this.

00:57:56.091 --> 00:58:03.191
Okay, so if you look across the cortex,

00:58:03.631 --> 00:58:10.831
the place where you see sort of the most relative diversity in gene expression

00:58:10.831 --> 00:58:15.211
are also the ones that are the, not the ancient parts of cortex,

00:58:15.411 --> 00:58:22.111
but the historically homologous parts of cortex that you can see in the same animals all the time.

00:58:22.111 --> 00:58:27.231
So you can, this is Leah Kruber's stuff, so you can always find a primary visual cortex.

00:58:27.371 --> 00:58:31.811
You can always find a primary somatosensory and an auditory cortex in all the mammals.

00:58:33.011 --> 00:58:38.351
And then if you look at, and so these identifiable regions have had the sort

00:58:38.351 --> 00:58:44.491
of most time to kind of accrue specific genetic information, okay? Yeah.

00:58:46.578 --> 00:58:51.698
And you see the most diversity in gene expression in them compared to the others.

00:58:52.718 --> 00:58:57.638
But the thing I think you pointed out, I put out in the lecture was,

00:58:57.778 --> 00:59:04.658
okay, so tell me something in the genetics of primary visual cortex that requires,

00:59:04.958 --> 00:59:09.098
that makes it optimal for being visual cortex other than getting visual information.

00:59:10.158 --> 00:59:16.558
Now, there may well be something, but I have been asking people for a long time

00:59:16.558 --> 00:59:20.578
now, and not in a hostile manner, because I really would like to know,

00:59:20.838 --> 00:59:23.938
okay, if you're going to make this visual cortex,

00:59:24.138 --> 00:59:29.378
does that mean that it has the neurotransmitter that's just perfect for the

00:59:29.378 --> 00:59:33.018
neurotransmitter receptor systems that are just perfect for the normal time

00:59:33.018 --> 00:59:38.778
course of visual events or the axon spread or whatever?

00:59:38.778 --> 00:59:42.998
Whatever, something about that that would really tailor primary visual cortex

00:59:42.998 --> 00:59:45.138
to its end. So that's what you'd want to ask.

00:59:45.418 --> 00:59:55.938
And so far, we only have stuff that shows up in the cortex because it gets a certain kind of input.

00:59:56.098 --> 01:00:02.618
So you see all these structures that almost certainly have some real innate

01:00:02.618 --> 01:00:05.878
component, but grow as a result of learning and experience.

01:00:06.258 --> 01:00:12.618
But we don't know So if there's anything in vision or somesthesis or audition

01:00:12.618 --> 01:00:17.838
or something that's specific to those regions and that makes the analysis of

01:00:17.838 --> 01:00:22.898
that kind of sensory information better because they typically end up in that particular place.

01:00:23.298 --> 01:00:30.778
Or maybe we are also biased in trying to interpret these areas too strongly in unimodal terms.

01:00:31.198 --> 01:00:36.798
Because you might find also multimodal responses in a visual area. And you certainly do.

01:00:37.818 --> 01:00:41.138
And that's the other half of this thing.

01:00:41.198 --> 01:00:48.138
You see so much stuff just recently about conversion of visual cortex into other

01:00:48.138 --> 01:00:52.298
uses in reading Braille or echolocation or whatever.

01:00:52.298 --> 01:00:57.338
Whatever, and in cases where you only, where you don't have to be blind from

01:00:57.338 --> 01:01:00.498
birth, but can just try to take up a braille hobby kind of recently.

01:01:01.118 --> 01:01:06.318
And the one interesting thing is that we're just sort of fixated on visual cortex.

01:01:06.458 --> 01:01:13.518
I've never seen anyone try to do anything similar, you know, does, you know,

01:01:15.628 --> 01:01:18.508
Do people who've lost sensation in

01:01:18.508 --> 01:01:22.228
their right hand use that for better understanding movies? I don't know.

01:01:22.908 --> 01:01:27.868
It's just that we don't tend to think of it as a surface that can be invaded

01:01:27.868 --> 01:01:29.888
so much or something. Right.

01:01:32.268 --> 01:01:36.008
So now we look very much at if you want the developmental program.

01:01:37.268 --> 01:01:41.628
But you also emphasized very much the role of motivation and embodiment.

01:01:42.488 --> 01:01:47.908
So how do those factors then really come in? To the development and the creation of a brain.

01:01:48.868 --> 01:01:56.448
Yeah, so, well, I originally started doing this research because I wanted to

01:01:56.448 --> 01:02:02.308
find out how you get brains wired for adaptive ends.

01:02:02.428 --> 01:02:09.008
And I wanted to find out, okay, if I wanted to be a really visual animal, how would I set that up?

01:02:10.588 --> 01:02:18.148
And what all this research taught me is that my initial guess about how to do

01:02:18.148 --> 01:02:19.648
that was entirely wrong.

01:02:19.828 --> 01:02:23.708
I imagine that you sort of somehow had a way of genetically identifying all

01:02:23.708 --> 01:02:32.908
the parts of the brain and body that were visual and you could somehow name

01:02:32.908 --> 01:02:39.608
them genetically and cause them to co-vary and that's how you would do that.

01:02:39.668 --> 01:02:42.948
Now I think that you generate this rather.

01:02:45.068 --> 01:02:49.488
Determinate in structure but plastic in content,

01:02:50.748 --> 01:02:59.148
brain and what you do to make a more visual brain is make the animal pay attention

01:02:59.148 --> 01:03:05.088
to its visual system particularly as for example we like to look at eyes and faces places,

01:03:05.088 --> 01:03:08.268
spend a lot of time learning about that,

01:03:08.368 --> 01:03:14.448
and then that's what the environmental loop in that, then that is what your brain comes to analyze.

01:03:14.888 --> 01:03:21.448
So I think if you look at where things really change in the brain from species

01:03:21.448 --> 01:03:26.528
to species, it's in this sort of basal forebrain, what motivational system is

01:03:26.528 --> 01:03:29.288
attached to those fundamental reinforcement circuitry.

01:03:29.288 --> 01:03:34.028
And if you're going to send an organism on a different path,

01:03:34.308 --> 01:03:36.648
you change what it cares about.

01:03:38.388 --> 01:03:44.108
And that's what I'm interested in looking at now as I think the central place

01:03:44.108 --> 01:03:49.088
where sort of the organism and the environment come together to specify what the brain consists of.

01:03:51.476 --> 01:03:59.256
And one of the things that I got from your talk was this notion of the adaptable

01:03:59.256 --> 01:04:05.336
nature of the vertebrate mammalian brain due to its kind of latent capacity.

01:04:05.636 --> 01:04:12.816
You know that we've been through this, what is it, 400 million year history of different species.

01:04:13.196 --> 01:04:18.816
And that in some way, modern brains have retained some of that history,

01:04:18.816 --> 01:04:24.156
even though you're adapted to some particular environment now,

01:04:24.976 --> 01:04:27.696
your ancestors were adapted to very different environments.

01:04:27.836 --> 01:04:34.136
And you gave the example of monkeys that have adapted to nocturnal living and

01:04:34.136 --> 01:04:40.256
have, I presume, fairly quickly evolved a lot of nocturnal visual capacities

01:04:40.256 --> 01:04:41.996
that you might see in other mammals.

01:04:42.276 --> 01:04:48.936
So am I right in understanding that as being quite a strong claim about the

01:04:48.936 --> 01:04:56.476
latent capability of the nervous system to recover these capabilities they've

01:04:56.476 --> 01:04:59.076
had in the past and roll them out when the opportunity arises.

01:04:59.976 --> 01:05:04.456
I wouldn't go quite that far. So I was making two kinds of claims there.

01:05:04.516 --> 01:05:09.176
One is for things that have been encountered routinely,

01:05:09.416 --> 01:05:19.876
like nocturnal versus diurnal, or olfactory is more useful information to me than visual,

01:05:20.036 --> 01:05:27.856
or the one that just came up as we were talking here, that this environment

01:05:27.856 --> 01:05:30.976
is really stable, this environment is not the sort of stress dimension.

01:05:32.996 --> 01:05:39.336
Then in those cases, you may well have the ability sort of retained to rather

01:05:39.336 --> 01:05:44.076
rapidly and in a coordinated way switch from one mode to another.

01:05:44.596 --> 01:05:47.876
Would that be sort of epigenetic in part?

01:05:48.056 --> 01:05:50.076
Sometimes epigenetic, but the

01:05:50.076 --> 01:05:54.436
ones I were talking about, none of them were epigenetic to my knowledge.

01:05:54.636 --> 01:05:59.576
Do you have an example in mammals of sort of epigenetic changes?

01:06:00.116 --> 01:06:06.176
Well, in the stress kinds of things where a particular kind of early environment

01:06:06.176 --> 01:06:10.936
is going to send you in a completely different direction for what kind of things

01:06:10.936 --> 01:06:13.636
you attend to and what motivates you and so forth.

01:06:14.585 --> 01:06:18.245
But that's not what I do my work on.

01:06:18.865 --> 01:06:27.265
But then the notion that the rest of the plasticity of the brain somehow embodies

01:06:27.265 --> 01:06:31.985
every possible thing that an ancestor has done, no. I don't think so.

01:06:32.965 --> 01:06:35.125
I don't think I've said quite that. That's a bridge too far.

01:06:36.725 --> 01:06:41.445
But there's a lot of what used to be called junk DNA. Now people think,

01:06:41.445 --> 01:06:44.245
well, actually it's got all this latent potential in it.

01:06:44.585 --> 01:06:47.685
And that explains the probably rapid transition.

01:06:48.365 --> 01:06:50.925
It's quite possible. It wouldn't be the first time.

01:06:52.865 --> 01:06:58.845
I have had to have been so careful talking about anything that vaguely sounds

01:06:58.845 --> 01:07:02.945
like that with the way biology has been and sort of adaptation.

01:07:02.945 --> 01:07:08.685
The notion that there could be anything other than the current adaptation state

01:07:08.685 --> 01:07:14.925
was such an evil thing to say for quite some time that it's taken me a while

01:07:14.925 --> 01:07:18.025
to get comfortable with even saying something like that out loud.

01:07:19.965 --> 01:07:26.845
Certainly in popular culture there's now this this this dockean view on evolution

01:07:26.845 --> 01:07:33.225
which just means it's very much sort of feet forward controlled by sort of these

01:07:33.225 --> 01:07:37.905
little fragments of DNA to dictate what the phenotype will look like.

01:07:38.045 --> 01:07:43.145
But now in what you're proposing it looks like the picture is becoming more

01:07:43.145 --> 01:07:49.205
complicated. So, do you see this really as a drastic departure from this more

01:07:49.205 --> 01:07:52.825
old-fashioned or old-fashioned, the traditional reductionist view?

01:07:53.145 --> 01:07:56.685
Or is it sort of an amendment of it?

01:07:58.045 --> 01:08:01.365
I don't really... It's the whole...

01:08:02.531 --> 01:08:09.231
The deviation that biology took to looking at only genes is the carrier of information

01:08:09.231 --> 01:08:13.211
at this point strikes me as just strange science.

01:08:13.431 --> 01:08:20.151
So, you know, so molded levels selection, for example, somehow that would be

01:08:20.151 --> 01:08:24.411
to me like saying, well, I can say that the properties of molecules,

01:08:24.591 --> 01:08:31.511
but I can never, ever describe the properties of gases, you know, which is just, you know.

01:08:31.511 --> 01:08:38.891
So collectively, you can talk about, in the kind of statistical language that

01:08:38.891 --> 01:08:44.511
I'm often using, there's variance that's accounted for at the level of species,

01:08:44.831 --> 01:08:49.511
there's at the level of taxon, there's all kinds of, it's the variation,

01:08:49.891 --> 01:08:57.611
useful variation across animals in anything you want to measure is not attached to the gene.

01:08:57.611 --> 01:09:03.251
And this is not a question about

01:09:03.251 --> 01:09:08.111
the usual, I mean, this is one version of multi-level selection arguments.

01:09:08.271 --> 01:09:13.331
There's one thing is all the causal structure at the level of selection for

01:09:13.331 --> 01:09:21.611
particular genes. But the second is just a more, you know, generic one.

01:09:22.131 --> 01:09:28.051
Can I use species to account for, for example, the relative,

01:09:28.311 --> 01:09:33.051
you know, predominance of different genes on Earth?

01:09:33.051 --> 01:09:37.111
Well, I'm sure the snowy owl would say that the presence of human genes on Earth

01:09:37.111 --> 01:09:40.171
has some sort of consequences for its frequency.

01:09:40.991 --> 01:09:47.011
And that's the kind of analysis of variance approach to the genome that just

01:09:47.011 --> 01:09:53.651
doesn't really get thought of as people have been talking about how you talk

01:09:53.651 --> 01:09:55.051
about causation in biology.

01:09:55.051 --> 01:10:02.951
So, people aren't used to considering the different levels of selection too

01:10:02.951 --> 01:10:04.791
much. And I think people will get better at it.

01:10:05.791 --> 01:10:12.371
So, Barbara, you're in this, the study of evolution and the brain for quite a while.

01:10:12.711 --> 01:10:16.531
And in some sense, you also have been changing your perspectives on this.

01:10:16.691 --> 01:10:23.051
Totally. So, if we want to follow in your footsteps, what's Barbara's law that we should adhere to?

01:10:25.051 --> 01:10:25.311
Um

01:10:27.666 --> 01:10:39.686
And there's a particular sort of unsettled or questioning state,

01:10:39.826 --> 01:10:44.286
which you can either ignore or you can attend to.

01:10:44.706 --> 01:10:49.806
And someone just told me a story, but I'm not really happy about it.

01:10:50.646 --> 01:10:55.326
And whenever you get that, I'm not really happy about it, pay attention to that.

01:10:55.326 --> 01:11:06.906
And try to figure out what the causes of why that story seems inadequate really put this major,

01:11:07.186 --> 01:11:15.366
you know, heavy alert system on in your head for that particular kind of gut feeling, essentially,

01:11:16.086 --> 01:11:17.986
that something is wrong with an explanation.

01:11:18.426 --> 01:11:21.566
Right. So pay attention to annoying surprises. Yes.

01:11:22.526 --> 01:11:26.626
So now five years from now, Tony and I are going to come and visit you at Cornell

01:11:26.626 --> 01:11:34.466
University to check whether you've been able to verify a prediction you're going to make today.

01:11:35.286 --> 01:11:40.226
So what's the most important prediction that you want to see tested in this

01:11:40.226 --> 01:11:42.046
time frame of about five years?

01:11:47.706 --> 01:11:52.946
Uh, that particular way of phrasing the question, I just, I need to rephrase

01:11:52.946 --> 01:11:55.066
a little because I like to,

01:11:55.426 --> 01:12:03.226
well, my prediction is that understanding motivational circuitry and its changes

01:12:03.226 --> 01:12:08.906
will be the way to understand how species differences emerge.

01:12:08.906 --> 01:12:15.426
But what we have is a complete absence of information about a lot of that.

01:12:15.946 --> 01:12:21.446
So people have really just begun. And I really think that for a lot of this

01:12:21.446 --> 01:12:29.626
kind of biology in general, we get into hypothesis testing way too soon.

01:12:29.726 --> 01:12:34.166
And the first thing that you need to do is describe the state of variation that's there.

01:12:34.806 --> 01:12:42.946
And so I'd be happy if I knew a lot more about what the actual variation was

01:12:42.946 --> 01:12:46.946
between species in there, how their motivational systems would be hooked up,

01:12:47.066 --> 01:12:49.686
and then we'll worry about predicting a little bit later.

01:12:49.966 --> 01:12:54.166
Okay, very good. Barbara Finley, thank you very much for this conversation. Okay, thank you.

01:12:54.626 --> 01:12:57.426
Thank you, that was great.

01:12:58.526 --> 01:13:04.386
The CSN podcast was produced by the Convergent Science Network of Biometrics

01:13:04.386 --> 01:13:10.746
and Biohybrid Systems, a project funded by the European Sevens Research Framework Program.

01:13:12.346 --> 01:13:17.646
For more interviews, recorded lectures, or upcoming conferences in the field

01:13:17.646 --> 01:13:23.906
of biometrics and biohybrid systems, go to csnnetwork.eu.

01:13:24.560 --> 01:13:33.520
Music.