WEBVTT

00:00:00.017 --> 00:00:06.037
I tried to learn for a while but I didn't get the dopamine that I needed,

00:00:07.377 --> 00:00:10.957
sex and drugs and rock and roll and I don't have milk, that's what you want,

00:00:12.277 --> 00:00:20.437
I should have yeah, I used artificial dopamine I did some of that that's what

00:00:20.437 --> 00:00:25.637
Eric said this is the Convergent Science Network podcast,

00:00:27.757 --> 00:00:31.397
leading researchers in the domain of neuroscience, brain theory,

00:00:31.557 --> 00:00:35.937
and technology are interviewed by Paul Vershoor and Tony Prescott.

00:00:39.917 --> 00:00:45.737
This is Paul Vershoor with the Convergent Science Network podcast with my colleague Tony Prescott.

00:00:46.237 --> 00:00:50.777
And we're talking with Greg Reckenzone, who was speaking today at the BCBT Summer

00:00:50.777 --> 00:00:55.937
School 2015 here in Barcelona, on something that I was told as a student didn't

00:00:55.937 --> 00:00:59.337
exist, which was adult cortical plasticity.

00:01:00.837 --> 00:01:06.477
So, Greg, why did you go over that topic? What's important about that question? question.

00:01:07.517 --> 00:01:12.237
Well, part of the reason why I did this, that I talked about this,

00:01:12.317 --> 00:01:17.657
which I don't usually talk about when I give seminars, is that Leah specifically asked me to.

00:01:18.017 --> 00:01:22.917
And she reasoned that back in the days when we were graduate students,

00:01:23.197 --> 00:01:25.257
it was, as you said, impossible and it didn't occur.

00:01:25.517 --> 00:01:29.877
And it was a real revolution in the way that we thought about how the cerebral cortex worked.

00:01:30.417 --> 00:01:33.217
And it's been so well accepted at

00:01:33.217 --> 00:01:36.097
this point that students today just kind of take it for granted

00:01:36.097 --> 00:01:38.857
and don't really get ever taught kind of

00:01:38.857 --> 00:01:42.117
the classic papers and studies that were done in order

00:01:42.117 --> 00:01:45.297
to demonstrate it to the point where we all now you

00:01:45.297 --> 00:01:48.977
know believe it so that was my main motivation today and plus you know it's

00:01:48.977 --> 00:01:52.657
eight years of my life working on that thing so it's kind of nice to revisit

00:01:52.657 --> 00:01:57.817
it and and um still see some enthusiasm about that kind of stuff right but now

00:01:57.817 --> 00:02:04.197
how do you explain that transition historically that's a good question too it It seemed,

00:02:04.257 --> 00:02:06.477
you know, as I was going through the beginning of my talk,

00:02:06.577 --> 00:02:10.997
that back in the 1800s, everybody knew that it had to exist and it was trivial.

00:02:11.197 --> 00:02:16.657
And then for some reason, since nobody had been able to really demonstrate how

00:02:16.657 --> 00:02:20.877
the brain changed when you actually did learn something, and then Hubel and

00:02:20.877 --> 00:02:23.697
Wiesel said, look, it stops changing after a certain period in development,

00:02:23.977 --> 00:02:27.297
which made perfect sense, it just got kind of lost in that shuffle.

00:02:27.297 --> 00:02:32.037
And so it took a 10-year hiatus, really, where people just kind of thought critical

00:02:32.037 --> 00:02:34.937
period, critical period, critical period, and nothing about,

00:02:35.117 --> 00:02:38.297
wait a minute, how did I just learn about this? I'm an adult.

00:02:38.397 --> 00:02:40.797
My brain had to have changed somehow, right, for me to learn this.

00:02:40.897 --> 00:02:42.857
But it couldn't have, but it just, you know.

00:02:44.384 --> 00:02:48.664
I have no idea why people got stuck on that and coming from mercenek's lab where

00:02:48.664 --> 00:02:54.244
he never considered that as a you know a serious proposition for learning i

00:02:54.244 --> 00:02:58.784
would never got caught up in that mindset right so i i always uh you always

00:02:58.784 --> 00:03:03.904
believe your mentor that's always important right i recommend exactly the opposite to my students.

00:03:06.144 --> 00:03:11.564
But now but then i add to that disagree with them but give a better alternative Yeah, absolutely.

00:03:11.804 --> 00:03:17.724
But for you, this history starts with Mike Mertzenich and John Kass. Right.

00:03:18.004 --> 00:03:23.064
Very much, right? So what was the key insight, you think, that brought them

00:03:23.064 --> 00:03:26.984
to that point of reviving these ideas about adult plasticity in the cortex?

00:03:27.284 --> 00:03:30.224
Well, I've talked to both of them about this a number of times.

00:03:30.404 --> 00:03:33.464
Because it was remarkable. It was a stretch. And it was a struggle.

00:03:33.604 --> 00:03:36.504
Because many people did not want to believe those early papers.

00:03:36.504 --> 00:03:40.424
Versus the antigen amputation paper and the median nerve section.

00:03:40.544 --> 00:03:42.684
They wanted to think of unmasking and all these other things.

00:03:42.884 --> 00:03:46.964
But they're both pretty passionate scientists, and they both thought that it

00:03:46.964 --> 00:03:49.084
just made more sense that it worked that way, right?

00:03:49.184 --> 00:03:53.864
And I think they just believed, right? And Mike would say that all the time,

00:03:54.284 --> 00:03:57.424
that he just kind of believed that it had to be something like that,

00:03:57.444 --> 00:03:58.764
and he just wanted to show how it was done.

00:03:58.924 --> 00:04:02.984
And so those early studies I thought were brilliant in the sense that they weren't

00:04:02.984 --> 00:04:06.544
supposed to work, and every time they did something, they learned something new.

00:04:06.884 --> 00:04:11.684
And so it was a really exciting time for everybody in those labs to be doing

00:04:11.684 --> 00:04:12.684
those kinds of experiments.

00:04:13.024 --> 00:04:18.224
So can you describe one of these sort of early experiments that really brought the message home?

00:04:19.004 --> 00:04:25.604
Well, for me, the one that actually solidified in my mind what happened is when

00:04:25.604 --> 00:04:29.384
you look at the map of the hand in Area 3b in a monkey,

00:04:29.504 --> 00:04:35.824
which is really precise and elegant, and each finger is represented very cleanly

00:04:35.824 --> 00:04:40.664
and distinctly and from the wrist to the fingertips, it's all very topographic, right?

00:04:40.884 --> 00:04:46.744
And then the critical period genetic model would say that if you lost a finger,

00:04:46.884 --> 00:04:50.464
then that part of the brain just wouldn't have anything to respond to anymore, right?

00:04:50.584 --> 00:04:53.684
And what they did is ask the question, is that really what happens?

00:04:53.984 --> 00:04:58.944
And so what they found was that indeed the neurons that used to respond to the

00:04:58.944 --> 00:05:02.284
missing finger now responded to the other ones. And another key insight that

00:05:02.284 --> 00:05:04.924
I didn't talk about today that really made, um.

00:05:06.260 --> 00:05:09.980
Made them think about it differently is that it wasn't simply the case that

00:05:09.980 --> 00:05:13.320
the old map was there and then some new things happened also.

00:05:13.900 --> 00:05:17.800
The whole map looked like a four-fingered monkey. So there was the same kind

00:05:17.800 --> 00:05:20.800
of overlap and the same progression and the same topography,

00:05:20.960 --> 00:05:22.180
but just with the missing finger.

00:05:22.320 --> 00:05:26.900
And so that gave them the insights that what must be happening is that the receptive

00:05:26.900 --> 00:05:29.060
field of a particular neuron has to be dynamic.

00:05:29.400 --> 00:05:33.200
It has to be able to change over a very short time course, right?

00:05:33.260 --> 00:05:36.520
Probably, you know, in a dramatic thing like this over the course of two months,

00:05:36.580 --> 00:05:39.620
but probably does it over the course of several days, right?

00:05:39.680 --> 00:05:43.420
And so there was something in there that these neurons are always actively changing

00:05:43.420 --> 00:05:47.040
the weights of their synapses to see what parts of the skin that they should

00:05:47.040 --> 00:05:50.700
represent relative to what all their neighboring neurons are representing, right?

00:05:50.800 --> 00:05:53.000
And that's how you keep the topography.

00:05:53.280 --> 00:05:56.620
Now, it was really lucky they did that in area 3B because if you look at the

00:05:56.620 --> 00:06:01.020
visual cortex that Charlie Gilbert did, it doesn't work that way.

00:06:01.540 --> 00:06:06.240
They all stack up, right? And so it's not this ordered topography that you see

00:06:06.240 --> 00:06:08.260
in the somatosensory cortex.

00:06:08.520 --> 00:06:13.560
In the auditory cortex, if you do a cochlear lesion, essentially what happens

00:06:13.560 --> 00:06:18.600
is the cells that used to respond to the now lesion frequencies respond to the

00:06:18.600 --> 00:06:21.700
one that was spared, the edge of the lesion.

00:06:21.920 --> 00:06:24.800
So all three of them are very different. And so it was really fortunate that

00:06:24.800 --> 00:06:30.320
they did the 3B study because it made the most sense, right,

00:06:30.400 --> 00:06:34.000
at the time. But also in retrospect, it was the most effective choice.

00:06:34.300 --> 00:06:37.580
Yeah, in retrospect, they didn't know about the other two, right?

00:06:37.640 --> 00:06:40.100
Those came subsequently, so they lucked out, essentially.

00:06:40.460 --> 00:06:46.340
But now, when these amputation studies were done with these results, did they...

00:06:47.895 --> 00:06:52.695
What was the resistance that you faced in the community?

00:06:53.975 --> 00:06:57.135
Yeah, so personally, fortunately, I didn't face any of that.

00:06:57.295 --> 00:07:03.395
So Mike was a champion. He would accept anybody's invitation to go out and talk

00:07:03.395 --> 00:07:07.655
to their research group and give a spiel and then spend the next half hour defending

00:07:07.655 --> 00:07:10.635
it because people said, no, no, you can't change the brain like that.

00:07:10.695 --> 00:07:16.095
It's just unmasking or you just, you know, it's just something that's not in

00:07:16.095 --> 00:07:19.155
any way related to any sort of perceptual differences, right?

00:07:20.075 --> 00:07:22.995
And I didn't see him a whole lot as a graduate student because he seemed to

00:07:22.995 --> 00:07:27.015
be gone all the time, going and just going out there and pitching his story

00:07:27.015 --> 00:07:30.315
and defending it the way that only he can do.

00:07:32.015 --> 00:07:34.795
And eventually people came around, I guess.

00:07:34.895 --> 00:07:38.315
I mean, when the accumulation of evidence, if he had stopped after the digital

00:07:38.315 --> 00:07:40.435
amputation, it would have been all over.

00:07:40.515 --> 00:07:43.275
That would have been the end of it, right? But we kept going and going and going.

00:07:43.295 --> 00:07:46.155
Then other people get on board too and start doing these experiments.

00:07:46.155 --> 00:07:50.075
And it's like, look, it always changes, right? So it must be real, right?

00:07:50.975 --> 00:07:58.615
So what is the now consensus? Because obviously we have a revised view of plasticity.

00:07:58.655 --> 00:08:03.235
This battle has been won, but also the notion of critical periods is still there.

00:08:03.375 --> 00:08:05.855
You know, we've talked about it several times in the last week.

00:08:06.095 --> 00:08:12.215
What would you see as the consensus now about the amount of adult plasticity

00:08:12.215 --> 00:08:15.835
and how that compares to plasticity in infancy.

00:08:16.595 --> 00:08:19.335
Well, there's absolutely no doubt that there's critical periods,

00:08:19.415 --> 00:08:23.635
and the lid sutures do cause ocular dominance column shifts and all these kinds

00:08:23.635 --> 00:08:25.615
of things. So that's absolutely certain, right?

00:08:26.495 --> 00:08:30.495
The way I think the field has come to a consensus is you have these developmental

00:08:30.495 --> 00:08:34.835
time points where you can do largely anatomical differences,

00:08:35.235 --> 00:08:36.195
right, but not necessarily.

00:08:36.555 --> 00:08:39.735
And then once all those things are over with, you're left

00:08:39.735 --> 00:08:42.895
with the adult plasticity

00:08:42.895 --> 00:08:46.135
which is how you change the weights of your synapses and

00:08:46.135 --> 00:08:48.975
how you mess with inhibition and things like that and that is

00:08:48.975 --> 00:08:54.135
what's basically the difference between critical periods and adult plasticity

00:08:54.135 --> 00:08:58.475
is an adult plasticity you don't expect much of any changes in the underlying

00:08:58.475 --> 00:09:03.175
anatomy so this fundamentally different mechanism yeah so it's fundamentally

00:09:03.175 --> 00:09:07.255
different and they're both working independently essentially.

00:09:09.255 --> 00:09:14.135
So now we look really at changing the periphery.

00:09:14.135 --> 00:09:19.435
We remove a finger or you switch your two fingers together to make a four-fingered

00:09:19.435 --> 00:09:23.955
monkey, and then you see these changes in the somatosensory representation of

00:09:23.955 --> 00:09:24.855
the body as you described.

00:09:27.055 --> 00:09:30.615
But now how far do these changes percolate into the system?

00:09:30.755 --> 00:09:35.495
Like, for instance, if I have a somatosensory change due to this change of my

00:09:35.495 --> 00:09:40.575
hand, And is it matched also to changes in, let's say, primary motor cortex, premotor cortex?

00:09:40.715 --> 00:09:44.415
How far do these changes really percolate down into the system?

00:09:46.059 --> 00:09:51.459
You know, that is a question that, to my knowledge, hasn't ever been looked at strongly.

00:09:51.719 --> 00:09:56.719
So I know that Randy Nudo did a whole series of experiments where he had the

00:09:56.719 --> 00:10:01.819
monkeys do different sorts of motor tasks, and then he looks at motor cortex,

00:10:01.839 --> 00:10:05.639
and there's predictable kinds of changes that would associate with that.

00:10:07.079 --> 00:10:14.239
And so clearly, any time you're interconnected networks, like 3B and 3A and

00:10:14.239 --> 00:10:18.459
M1, one, you're going to see these changes percolate at least to the extent

00:10:18.459 --> 00:10:19.819
that they need to be, right?

00:10:19.939 --> 00:10:24.379
So I imagine that if I looked in the motor cortex of those same animals that

00:10:24.379 --> 00:10:28.599
I studied, I would see all kinds of things that were related to the reach and

00:10:28.599 --> 00:10:30.979
to the release and all these things because they were doing that,

00:10:31.019 --> 00:10:32.699
and that was very important to them too, right?

00:10:32.959 --> 00:10:36.819
So is that because of the somatosensory changes? Probably not as much as it

00:10:36.819 --> 00:10:39.099
is to the actual motor movement, but they're going to work together.

00:10:39.819 --> 00:10:42.799
But auditory cortex didn't change, even though they heard this stuff,

00:10:42.959 --> 00:10:45.639
right? So there's some limit to how much it goes.

00:10:45.759 --> 00:10:48.939
And it's probably just to what's really primarily task-related.

00:10:49.619 --> 00:10:52.499
Because that was sort of the next step in your experiments, right?

00:10:52.939 --> 00:10:58.879
So now, first we change the periphery. You could say, okay, it's a rather strong

00:10:58.879 --> 00:11:01.399
perturbation, so we have a reorganization of the map.

00:11:01.739 --> 00:11:06.139
But then you start to look really at the learning component of this by engineering

00:11:06.139 --> 00:11:12.659
the task where you provide haptic feedback back or tactile feedback to the, to the monkey.

00:11:13.719 --> 00:11:18.259
Um, and then trying to see what kind of changes you would then get in,

00:11:18.279 --> 00:11:20.479
in, uh, the smetacensory cortex.

00:11:20.759 --> 00:11:24.999
So, um, what kind of changes did you get there in that experiment?

00:11:26.209 --> 00:11:30.989
The changes in the somatosensory? Okay, so the monkeys were trained to feel

00:11:30.989 --> 00:11:34.509
a vibration on one small part of their finger, starting at 20 hertz,

00:11:34.589 --> 00:11:38.329
and to detect when it went faster in a subsequent presentation.

00:11:38.489 --> 00:11:42.269
So we saw a great deal of changes, some of which were related to the behavior,

00:11:42.329 --> 00:11:43.669
and many of which were not.

00:11:43.829 --> 00:11:47.409
Okay, so one thing that wasn't related to the behavior, as far as we could tell,

00:11:47.589 --> 00:11:49.689
was the size of the receptive fields got bigger.

00:11:50.009 --> 00:11:53.889
And the thinking at the time then was that, well, that's because it's not,

00:11:53.889 --> 00:11:55.209
It's a pretty big stimulus.

00:11:55.349 --> 00:11:59.029
It vibrated a lot of skin, and so everything is being synchronously activated

00:11:59.029 --> 00:12:04.709
by this 20-hertz stimulus, and that's a consequence of the mechanisms by which

00:12:04.709 --> 00:12:06.149
you change your weights.

00:12:06.149 --> 00:12:12.029
Another thing that happened was there was an expansion of the representation of that skin.

00:12:12.989 --> 00:12:18.389
And that was correlated with the ability of the monkey to do the task, but not super well.

00:12:18.589 --> 00:12:22.429
And if that had been the end of the day, I would have been perfectly happy with

00:12:22.429 --> 00:12:25.389
it, because that's what I was looking for, a change in the map related in some

00:12:25.389 --> 00:12:26.469
way to the change in the behavior.

00:12:27.009 --> 00:12:31.349
The main thing that was related to the behavior was the ability of the neurons

00:12:31.349 --> 00:12:36.009
to follow each individual cycle of that 20 hertz or 21 or 24 hertz stimulus.

00:12:36.389 --> 00:12:41.249
And what was interesting to me, the two changes that really made the big difference

00:12:41.249 --> 00:12:48.049
was for the brain, when the skin was stimulated on the finger that was being trained,

00:12:48.229 --> 00:12:52.369
there was a lot more neurons that responded in that way to each cycle of the stimulus.

00:12:53.109 --> 00:12:58.789
But every individual neuron that did respond that way responded about as well

00:12:58.789 --> 00:13:01.249
as ones that did the untrained skin.

00:13:01.509 --> 00:13:06.709
So if you did some sort of measurement of the fidelity of the response relative

00:13:06.709 --> 00:13:10.769
to the stimulus itself, there was no difference between trained and untrained.

00:13:12.109 --> 00:13:16.529
And so that would mean, well, the signal is just louder because there's more neurons doing it.

00:13:16.589 --> 00:13:19.709
But it was more than that. What happened that really made the difference was,

00:13:20.569 --> 00:13:25.249
for the untrained skin or normal skin, if you will, So the way that the timing

00:13:25.249 --> 00:13:30.309
of the response to each cycle of the stimulus varied quite a bit between individual neurons.

00:13:30.689 --> 00:13:35.209
So that would make kind of, if you did all the population, you have a broad tuning function.

00:13:35.469 --> 00:13:38.569
And the trained monkeys, they all lined up with each other.

00:13:38.769 --> 00:13:42.009
So the temporal fidelity across the population was enhanced.

00:13:42.149 --> 00:13:43.989
So it was a louder signal, it was a better signal.

00:13:44.649 --> 00:13:48.649
So it got much tighter, which allowed the monkey, I'm interpreting,

00:13:48.909 --> 00:13:52.689
to be better able to tell the difference between 20 and 22 hertz because there

00:13:52.689 --> 00:13:56.149
was less slop, less variability, and it was a much cleaner, tighter signal.

00:13:56.569 --> 00:14:01.609
But you would say, is that expressing something like what fires together wires together?

00:14:01.789 --> 00:14:05.449
Like I'm driving these neurons simultaneously, they fire simultaneously,

00:14:05.949 --> 00:14:08.829
this drives local plasticity, they get coupled more tightly together,

00:14:09.069 --> 00:14:14.269
and as a result, their response becomes sharper in time. Absolutely. Okay.

00:14:15.038 --> 00:14:18.158
But then the other thing you showed is all this elongation of these receptive

00:14:18.158 --> 00:14:20.978
fields. Right. That's the thing that didn't make any sense to me.

00:14:21.598 --> 00:14:24.958
Why would these receptive fields get so much larger in terms of their sensitivity

00:14:24.958 --> 00:14:28.398
along the finger? Along the finger.

00:14:28.898 --> 00:14:34.758
Yeah, so when the monkey has his hands on the apparatus and the thing comes

00:14:34.758 --> 00:14:37.578
up and starts to vibrate, it vibrated quite a bit, right?

00:14:37.838 --> 00:14:41.638
And so it wasn't the same as when you're doing the mapping experiment and you're

00:14:41.638 --> 00:14:45.098
just barely touching the skin. It's actually a pretty significant stimulus.

00:14:45.278 --> 00:14:47.458
And I felt it. I wanted to see what it felt like.

00:14:47.698 --> 00:14:53.418
And it felt pretty big, right? So it was as though, even though the probe was

00:14:53.418 --> 00:14:56.078
kind of small, since it was moving so much, it was moving a lot of skin,

00:14:56.138 --> 00:14:57.058
almost the whole phalanx.

00:14:57.538 --> 00:15:01.838
And I think that's wires, fires together, wires together, is the same kind of

00:15:01.838 --> 00:15:04.078
thing. You're moving all the skin together at once, right?

00:15:04.398 --> 00:15:07.498
And there was no penalty for having a big receptive field, right?

00:15:07.598 --> 00:15:11.838
I would guess that if I had two probes and I asked the monkey to tell me,

00:15:11.858 --> 00:15:14.858
am I just doing one or two then they would get small but

00:15:14.858 --> 00:15:17.778
it needs more than files together wires together because there's

00:15:17.778 --> 00:15:21.878
an effective reinforcement here that you need the reinforcement right

00:15:21.878 --> 00:15:25.058
right exactly they have to they have to know that they're

00:15:25.058 --> 00:15:28.158
doing it right and it's all this making the script discrimination

00:15:28.158 --> 00:15:31.458
gets you the peanut whatever right exactly yeah

00:15:31.458 --> 00:15:34.578
so it's a more complicated system than simply

00:15:34.578 --> 00:15:37.358
heavy and learning you know if you experience something a lot

00:15:37.358 --> 00:15:40.818
then through heavy in learning you might refine your receptors

00:15:40.818 --> 00:15:43.718
but if right if it's important for a reward then that's

00:15:43.718 --> 00:15:46.638
going to be i think as you've shown much bigger effect much bigger

00:15:46.638 --> 00:15:50.678
effect so if you do exactly the same stimulation but the monkey listens to the

00:15:50.678 --> 00:15:54.298
sounds instead then you don't see the big receptor fields in the big area and

00:15:54.298 --> 00:15:59.818
the tighter tuning so this this is this is not heavy and then or at least it's

00:15:59.818 --> 00:16:03.938
a heavy role that's being modulated well that's the question right Is it just

00:16:03.938 --> 00:16:05.498
driven by the statistics of the input,

00:16:05.758 --> 00:16:10.558
or does it depend on an additional neuromodulatory signal that says,

00:16:10.758 --> 00:16:12.798
ah, this is really great, we should get it again?

00:16:13.138 --> 00:16:15.198
Yeah, I think it's the latter. Okay. Yeah.

00:16:16.058 --> 00:16:21.038
But now, the other thing you saw is also in terms of the response latency,

00:16:21.238 --> 00:16:22.258
there were changes, right?

00:16:22.458 --> 00:16:30.798
The neurons also seem to be responding earlier to the stimulus than the control

00:16:30.798 --> 00:16:33.818
condition. Mm-hmm . So how would you count real effect?

00:16:35.023 --> 00:16:38.723
You know, that's a really good question, too. Why would you get,

00:16:38.883 --> 00:16:42.663
I mean, it's clear to see why you would get sharper, right?

00:16:42.743 --> 00:16:44.783
But why would you get sharper and earlier, too?

00:16:45.323 --> 00:16:48.783
And I have no really good answer for that, right?

00:16:48.903 --> 00:16:52.463
So it's somehow as though these early responses are the ones that the monkey

00:16:52.463 --> 00:16:57.083
was using and pulling the other neurons towards the shorter latency. Okay.

00:16:57.543 --> 00:17:00.843
But it was a discrimination test that the monkey had to perform.

00:17:01.183 --> 00:17:08.863
Right. Right? So you changed the stimulation frequency and the monkey had to

00:17:08.863 --> 00:17:09.983
say, ah, I detected that.

00:17:10.203 --> 00:17:13.063
Right. If it was correct, it got a reward. Right. Right?

00:17:13.663 --> 00:17:16.703
So how frequent were these changes in frequency?

00:17:17.763 --> 00:17:23.083
So the way we decided to do it is the first one was always 20 hertz.

00:17:23.083 --> 00:17:27.303
If I remember correctly, the second one was always 20 Hz, and then it could

00:17:27.303 --> 00:17:31.163
change on the third, fourth, fifth, sixth, or seventh with equal probability.

00:17:31.703 --> 00:17:37.323
So on average, there was five-ish of these 600 millisecond duration things.

00:17:37.463 --> 00:17:40.563
So we were thinking at the time that we had to do a lot of stimulation because

00:17:40.563 --> 00:17:44.883
the Jenkins experiment had millions of stimuli and saw big changes.

00:17:45.023 --> 00:17:48.623
So I was afraid that if you didn't do quite so many,

00:17:48.783 --> 00:17:52.283
if you did just one and then another

00:17:52.283 --> 00:17:55.243
other one you know two choice right then that wouldn't

00:17:55.243 --> 00:17:58.043
necessarily do it i think now that that would have worked and probably would

00:17:58.043 --> 00:18:00.683
have saved me a lot of time but right so i

00:18:00.683 --> 00:18:04.063
mean looking across these two experiments you see that differentiation

00:18:04.063 --> 00:18:07.023
leads to changes in the map that that

00:18:07.023 --> 00:18:10.503
probably have a lot to do just with the input

00:18:10.503 --> 00:18:14.323
statistics to the map because this is task there's no task dependent anyway

00:18:14.323 --> 00:18:19.683
right so well on when we started bringing a task and it's really very much also

00:18:19.683 --> 00:18:23.263
the reward content or the task contingency the relation to reward or punishment

00:18:23.263 --> 00:18:27.743
that would impose further changes to the map you you read that interpretation

00:18:27.743 --> 00:18:29.603
there's no learning mechanisms.

00:18:30.403 --> 00:18:33.223
At work okay but now how stable would

00:18:33.223 --> 00:18:36.363
that learning be like for instance if what you

00:18:36.363 --> 00:18:39.743
already saw what these these peripheral changes like fusing two

00:18:39.743 --> 00:18:44.703
fingers and then disconnecting them actually the map follows that manipulation

00:18:44.703 --> 00:18:51.703
but it might do it more slowly as it might follow let's say the task contingency

00:18:51.703 --> 00:18:56.183
in in the the stimulation experiment you described right so what is sort of

00:18:56.183 --> 00:18:59.723
the the characteristic time constants of these two learning processes,

00:19:00.463 --> 00:19:07.763
yeah that's something that um i never address specifically but it's obviously a critical,

00:19:08.483 --> 00:19:13.103
component mostly because the idea of course is that you can use this for learning

00:19:13.103 --> 00:19:16.123
for good for rehabilitation and all these kinds of things and how long does

00:19:16.123 --> 00:19:22.003
it take to do this and once you do change do you ever want to change back or

00:19:22.003 --> 00:19:24.923
can you change back or do you need to change back and.

00:19:26.453 --> 00:19:30.853
So, I think that's going to be really largely task-dependent and really largely

00:19:30.853 --> 00:19:36.393
how much neuromodulator can you get in there based on pharmaceutical or just

00:19:36.393 --> 00:19:41.493
the desire to do it or the importance of the stimulus to you, et cetera.

00:19:41.673 --> 00:19:47.153
So, my guess is that it will probably take days or weeks or months depending

00:19:47.153 --> 00:19:49.333
on all those different kinds of things, right?

00:19:49.473 --> 00:19:53.693
Okay. But in case of the differentiation or the amputation experiment,

00:19:54.113 --> 00:19:56.833
how much time does does it take for that map

00:19:56.833 --> 00:20:00.593
to stabilize yeah unknown okay so

00:20:00.593 --> 00:20:04.033
uh those early experiments they also did uh

00:20:04.033 --> 00:20:06.993
two-digit amputations and they waited

00:20:06.993 --> 00:20:09.973
two months or three months or whatever and what

00:20:09.973 --> 00:20:13.053
they found there was that there would be a silent zone so the

00:20:13.053 --> 00:20:18.813
two-digit amputation deafferent too big of a portion of the cortex for the afferents

00:20:18.813 --> 00:20:24.213
to actually reach each other then ted jones and And Tim Pons looked at the silver

00:20:24.213 --> 00:20:30.253
spring monkeys a decade or two later that had dorsal rhizotomies done to them

00:20:30.253 --> 00:20:32.273
15 or 20 years beforehand,

00:20:32.473 --> 00:20:35.473
and they saw complete recovery in the map.

00:20:35.553 --> 00:20:38.633
And that was correlated with changes in anatomical distributions as well,

00:20:38.673 --> 00:20:41.433
all the way from the cuneate and the thalamus up to the cortex.

00:20:41.713 --> 00:20:45.473
So if you wait long enough, presumably, there was a huge deafferentation.

00:20:46.133 --> 00:20:50.253
That's the whole arm, right? And they still saw centimeters of change as opposed

00:20:50.253 --> 00:20:54.353
to... I think the early studies were something like 500 or 600 microns.

00:20:54.353 --> 00:20:57.973
If you got beyond that, you couldn't recover or you couldn't change your map.

00:20:58.133 --> 00:21:00.613
But you know, you only wait two months, right? If you're not going to really

00:21:00.613 --> 00:21:04.853
wait 10 years, that's a hard grant to get in the United States.

00:21:06.113 --> 00:21:10.373
But sort of the task-dependent change under the influence of neuromodulation

00:21:10.373 --> 00:21:14.233
can go relatively rapid, right? Right. Months, yeah.

00:21:14.913 --> 00:21:19.533
Days or weeks, yeah. With a task-specific change, certainly if you go to the

00:21:19.533 --> 00:21:24.513
matter of conditioning, You might see MAP changes in dozens of trials, no? Oh, yeah, sure.

00:21:24.873 --> 00:21:27.873
Norm Weinberger has done that for decades. Exactly. Yeah. Right.

00:21:27.933 --> 00:21:31.293
Yeah, yeah. Well, then these amputation experiments show that the overall,

00:21:31.433 --> 00:21:34.953
let's say, topological organization of the MAP depended on the periphery.

00:21:34.953 --> 00:21:38.313
It might be a much slower process that might require, let's say,

00:21:38.353 --> 00:21:41.773
throwing out new processes, building new connections, and so on.

00:21:42.273 --> 00:21:44.353
Would you see it like this? That's it.

00:21:45.484 --> 00:21:49.744
The amputation, the processes underlying the reorganization of amputation are

00:21:49.744 --> 00:21:53.544
slow, and the task-specific ones, depending on neuromodulation,

00:21:53.544 --> 00:21:56.824
are fast, or you wouldn't really see it in those simple terms?

00:21:56.824 --> 00:22:01.344
I would say that's probably accurate, because when you think about a small deafferentation.

00:22:01.524 --> 00:22:09.484
The consequences of that to the animal are not as dire as you have to do this

00:22:09.484 --> 00:22:11.304
task in order to get your food.

00:22:11.304 --> 00:22:14.684
Right so so how much neuromodulation would

00:22:14.684 --> 00:22:17.564
there be in in a situation where you

00:22:17.564 --> 00:22:20.764
have a digit amputation it's probably not nearly as much so in

00:22:20.764 --> 00:22:23.704
that sense it would go much slower because the need for it

00:22:23.704 --> 00:22:27.644
to go faster is just not there you mentioned um

00:22:27.644 --> 00:22:30.684
in the talk you were talking about this is requiring attention

00:22:30.684 --> 00:22:33.664
uh and uh so what's interesting

00:22:33.664 --> 00:22:36.764
is attention doesn't necessarily imply reinforcement

00:22:36.764 --> 00:22:40.104
it implies some maybe more intrinsic motivation

00:22:40.104 --> 00:22:42.884
to attend i guess you're not directly getting

00:22:42.884 --> 00:22:45.964
at it in this task but uh for instance

00:22:45.964 --> 00:22:48.864
in the case of a person who's learning

00:22:48.864 --> 00:22:51.804
some motor skill their reinforcement might

00:22:51.804 --> 00:22:54.824
be you might have to wait a long time for it but uh

00:22:54.824 --> 00:22:57.644
you're intrinsically motivated so that right so we

00:22:57.644 --> 00:23:01.084
don't necessarily have to think about uh well maybe

00:23:01.084 --> 00:23:04.084
we do want to think about a person you're modulated but the the uh

00:23:04.084 --> 00:23:07.204
system that's delivering this could be some really

00:23:07.204 --> 00:23:10.584
quite complex uh system around intrinsic

00:23:10.584 --> 00:23:13.504
motivation signals rather than you know

00:23:13.504 --> 00:23:19.324
absolutely rewards right so imagine learning to play a guitar right which is

00:23:19.324 --> 00:23:23.444
difficult to do i can say from personal experience and the motivation is to

00:23:23.444 --> 00:23:28.324
sound good yeah right and and that's very esoteric right so now your auditory.

00:23:28.324 --> 00:23:31.544
Cortex is deciding how much attention or a modulator.

00:23:31.784 --> 00:23:35.564
Your motor system is going to get, and your tactile system based on.

00:23:36.184 --> 00:23:41.544
You know, your desire, whatever that is, to sound like, you know,

00:23:41.544 --> 00:23:44.644
not a hack, right? To sound good. So, yeah.

00:23:44.884 --> 00:23:47.964
And you don't necessarily know what the task is. I mean, it's,

00:23:47.964 --> 00:23:51.184
you know, you don't know what you need to do to sound good. You're exploring

00:23:51.184 --> 00:23:52.824
the space. There's a lot of trial and error.

00:23:53.064 --> 00:23:56.984
Right. But I guess you're getting at the basis for that here.

00:23:57.804 --> 00:24:02.424
Right. Right. But do we really need to have an attentional interpretation of this?

00:24:02.564 --> 00:24:07.864
I mean, if you just get this, the banana chow, whatever you give these animals

00:24:07.864 --> 00:24:17.144
as a reward, driving in neuromodulators like dopamine, which modulates plasticity in the cortex.

00:24:18.838 --> 00:24:21.698
Leading to this reorganization do we need to say

00:24:21.698 --> 00:24:24.718
oh and we paid attention to it do we really need that

00:24:24.718 --> 00:24:28.258
interpret it's kind of right so task contingency it

00:24:28.258 --> 00:24:33.018
could be task contingency it could be you know reward uh dopamine it could be

00:24:33.018 --> 00:24:36.778
norepinephrine it could be all kinds of different things so so to use the word

00:24:36.778 --> 00:24:41.638
attention uh as i did in my talk was a pretty loose interpretation of the people

00:24:41.638 --> 00:24:46.138
who'd study attention right right so they they would they would shy away from that

00:24:46.218 --> 00:24:48.538
necessarily being the key, right?

00:24:50.418 --> 00:24:56.118
I kind of throw the word out there just because right after I left Mike's lab,

00:24:56.198 --> 00:25:01.758
Mike Kilgard came and he essentially did the auditory version of the experiments in rats.

00:25:02.018 --> 00:25:05.698
And instead of training them, he just electrically stimulated the basal forebrain

00:25:05.698 --> 00:25:07.478
and got all the same effects.

00:25:07.658 --> 00:25:10.438
Right, exactly. And there was no attention there, right?

00:25:10.938 --> 00:25:14.538
But that was auditory and auditory conditioning, right? Right,

00:25:14.538 --> 00:25:15.358
exactly. Yeah, exactly.

00:25:15.618 --> 00:25:19.378
Yeah. So he dumps a bunch of acetylcholine on the cortex right when the sounds

00:25:19.378 --> 00:25:22.818
are playing and the maps change like crazy. Right. Exactly. Yeah.

00:25:23.498 --> 00:25:28.318
So then, so after these experiments that really demonstrate,

00:25:28.698 --> 00:25:32.598
that really contributed significantly to demonstrating that we have a dot plasticity

00:25:32.598 --> 00:25:36.318
in the cortex and to show also some of its organizing principles,

00:25:36.698 --> 00:25:39.498
in particular the role of neuromodulators, I think,

00:25:39.578 --> 00:25:45.298
and let's say input statistics on on how these maps are organized um what are

00:25:45.298 --> 00:25:49.618
the boundaries on that process though i mean would you be able if i would give

00:25:49.618 --> 00:25:54.678
you just enough time and monkeys and training equipment could you just easily

00:25:54.678 --> 00:25:57.298
retrain the auditory cortex into a visual cortex.

00:25:59.114 --> 00:26:03.174
And make it as large or small as you wanted it to be. In adults.

00:26:03.394 --> 00:26:06.574
Yeah. Take normal adults and then see how you can manipulate that.

00:26:07.434 --> 00:26:14.614
Yeah. So we heard a couple of days ago in the seminar that you blindfold a person

00:26:14.614 --> 00:26:19.034
and their visual cortex will start helping them learn Braille, right?

00:26:19.294 --> 00:26:22.674
So in that sense, you know. Or to echolocate.

00:26:22.814 --> 00:26:27.174
Or to echolocate, exactly. So in that sense, you know, all bets are off.

00:26:27.174 --> 00:26:31.614
It sounded like from that limited amount of data right but where would you start

00:26:31.614 --> 00:26:35.574
when we're going to make an echolocating monkey where would you start how would

00:26:35.574 --> 00:26:36.594
you do this most efficiently.

00:26:39.474 --> 00:26:42.354
Uh probably easier to teach him to read braille than to

00:26:42.354 --> 00:26:48.314
echolocate at least from the training point of view and then uh if if you were

00:26:48.314 --> 00:26:52.714
going to do that if you deprived a bunch of cortex of its normal input then

00:26:52.714 --> 00:26:57.674
it would and and demanded that it use its hands you know efficiently well and

00:26:57.674 --> 00:27:00.394
its visual cortex, for example, is not doing much of anything,

00:27:00.634 --> 00:27:05.374
I think it would take days to weeks to start getting that activated without

00:27:05.374 --> 00:27:11.454
doing the neonatal manipulations that you can do to get all this rewiring that

00:27:11.454 --> 00:27:13.394
Magranca showed a long time ago.

00:27:14.434 --> 00:27:19.754
So, yeah, you know, in 1992 when those publications came out,

00:27:19.814 --> 00:27:22.714
I'd say, well, that's never going to happen because there's no substrate for

00:27:22.714 --> 00:27:24.134
that, right? It's just not going to work.

00:27:24.194 --> 00:27:28.574
It's all cortical and it's all done, right? Today, we know much more about it,

00:27:28.614 --> 00:27:30.714
and so it is possible to do these kinds of things.

00:27:30.814 --> 00:27:35.414
And exactly how that's done is still not entirely clear, but it happens,

00:27:35.654 --> 00:27:38.174
so you can change these things.

00:27:38.974 --> 00:27:43.874
So the second part of the talk, you focused on sound localization, right?

00:27:43.914 --> 00:27:48.054
So we moved to different preparation, we moved also to different kinds of tasks.

00:27:48.274 --> 00:27:53.634
Right. So why does sound localization help you to understand this issue of adult

00:27:53.634 --> 00:27:56.554
plasticity? Great, so um.

00:27:57.989 --> 00:28:00.829
I started doing those sound localization experiments when I became an assistant

00:28:00.829 --> 00:28:03.649
professor at least five or more years ago.

00:28:04.509 --> 00:28:14.589
And really my motivation at that time was to learn what is it about the brain

00:28:14.589 --> 00:28:16.589
that allows us to perceive complicated things.

00:28:17.749 --> 00:28:21.489
And the plasticity part of my life, it seemed to me, was pretty much over by

00:28:21.489 --> 00:28:24.069
then. So I came, I conquered, I left, right?

00:28:24.129 --> 00:28:28.169
And I wanted to make a name for myself independent of Mike Merzenich and Bob

00:28:28.169 --> 00:28:32.289
Wirtz, which is no easy task because these guys are just fantastic scientists, right?

00:28:32.609 --> 00:28:37.729
And so, like I said in the talk, there was really not much going on in the auditory cortex.

00:28:38.029 --> 00:28:41.809
And I think it's really, really important, right?

00:28:42.309 --> 00:28:46.009
So human beings, a lot of people will say, a human being is a visual animal, right?

00:28:46.069 --> 00:28:50.109
So we have this huge visual cortex and we just like to look at stuff,

00:28:50.269 --> 00:28:52.969
right? The other argument is that we're really a social animal,

00:28:53.129 --> 00:28:57.289
and what we really want to do is talk to each other.

00:28:57.709 --> 00:29:00.669
That's a strong motivator in the human brain.

00:29:01.129 --> 00:29:06.529
If you can't talk to other people, usually you go crazy.

00:29:07.069 --> 00:29:10.009
You really need that sort of social interaction. actions. So in that sense,

00:29:10.029 --> 00:29:13.589
we have this great auditory system that lets us do things like understand speech.

00:29:14.289 --> 00:29:17.589
And so in that sense, I thought, well, this is equally, if not more important

00:29:17.589 --> 00:29:21.969
than visual system, right? So it should get some people studying it.

00:29:22.049 --> 00:29:25.989
And we have the visual system to compare our studies to.

00:29:26.129 --> 00:29:30.069
So on the one hand, if you do the same kinds of studies that have been done

00:29:30.069 --> 00:29:33.609
decades ago in the visual system and the auditory system, it's new, right?

00:29:33.709 --> 00:29:37.569
You already had this all sort of figured out if it works exactly the same way, right?

00:29:37.849 --> 00:29:42.129
So my motivation was to go ahead and do something that was novel.

00:29:42.309 --> 00:29:46.269
And what I really wanted to do was understand how complicated perceptions could

00:29:46.269 --> 00:29:47.849
be formed in the cerebral cortex.

00:29:48.049 --> 00:29:51.669
So people have said, oh, you were a somatosensory guy, then you were an auditory

00:29:51.669 --> 00:29:52.429
guy, then you were a vision guy.

00:29:52.529 --> 00:29:56.129
I've always been a cortex guy. So I'm a clinical snob, I'll admit it.

00:29:56.189 --> 00:30:00.189
And I've always been interested in how does the cortex drive perceptions, right?

00:30:00.409 --> 00:30:03.469
One way you can do it is you can have plasticity to get better.

00:30:03.629 --> 00:30:06.269
Another way you can do it is you say, well, how do you do it without a whole

00:30:06.269 --> 00:30:08.069
bunch of plasticity? How do you do this?

00:30:08.369 --> 00:30:11.969
And the auditory system and the sound localization system is really nice in

00:30:11.969 --> 00:30:16.149
my mind because it's not topographic and I know what it's trying to represent,

00:30:16.329 --> 00:30:17.949
which is azimuth and elevation.

00:30:18.229 --> 00:30:22.749
So I know exactly what the thing's trying to do and that will give me some handle

00:30:22.749 --> 00:30:24.689
on figuring out how it does that. Yeah.

00:30:26.077 --> 00:30:30.977
So not being topographic was a motivation you mentioned also in the talk. Right.

00:30:32.337 --> 00:30:37.357
But doesn't that make it life harder for you if you take something which is not topographic?

00:30:37.497 --> 00:30:42.277
Or is the goal to show that the same principles apply whether there's topography or not?

00:30:42.557 --> 00:30:47.437
Well, it does make it a lot harder, right? So if you're nicely topographic,

00:30:47.677 --> 00:30:53.197
you can do some really cool experiments like microstimulation or microlesions and things like this.

00:30:53.297 --> 00:30:55.617
And you can probe the system much more easily.

00:30:56.077 --> 00:31:00.817
I had tried for a while to do microstimulation of the auditory cortex to see

00:31:00.817 --> 00:31:01.917
if I could change the percept.

00:31:02.857 --> 00:31:05.497
Ken Britton is a colleague of mine at the Center for Neuroscience,

00:31:05.577 --> 00:31:09.057
and he has a long history of microstimulating MT and showing really elegantly

00:31:09.057 --> 00:31:11.677
that it does perturb the perceptual abilities of the animal.

00:31:12.057 --> 00:31:16.797
And he loaned me all his stimulation stuff, and nothing ever seemed to work.

00:31:16.917 --> 00:31:19.977
And I think it's likely because you can't just

00:31:19.977 --> 00:31:22.757
pick one spot like he could in

00:31:22.757 --> 00:31:25.597
MT with the best direction and set your stimulus up to

00:31:25.597 --> 00:31:29.457
do that instead you stimulate this one

00:31:29.457 --> 00:31:32.237
percent of the neurons that are contributing to the percept and

00:31:32.237 --> 00:31:35.757
you can never measure that in the noisy thing so

00:31:35.757 --> 00:31:40.017
there's a distinct disadvantage to that so the real motivation was thinking

00:31:40.017 --> 00:31:44.617
that um a lot of the brain a lot of the interesting parts of the brain that

00:31:44.617 --> 00:31:48.017
we do cognitive things in that we can't really study because we don't know anything

00:31:48.017 --> 00:31:52.557
about how those cells work because they're not topographic and it's not an easy label.

00:31:52.637 --> 00:31:55.997
If we understood how you could do it in the cortex and something that you know

00:31:55.997 --> 00:31:57.977
is not topographic, then maybe

00:31:57.977 --> 00:32:01.517
you'd get some insights into how to look at other places in the brain.

00:32:02.923 --> 00:32:08.103
But now that you say you want to understand perception and complex perception,

00:32:08.263 --> 00:32:11.483
we look at localization where in some sense you can say, well,

00:32:11.503 --> 00:32:14.883
all I need to do is just know the interaural time difference and the attenuation

00:32:14.883 --> 00:32:18.883
that I get through my pina and my skull, and I have a good estimate of where

00:32:18.883 --> 00:32:23.243
things are, and I can just delegate this down to my superior colliculus. Right.

00:32:23.503 --> 00:32:27.403
So what's the leverage looking at sound localization?

00:32:27.523 --> 00:32:29.603
Actually, it is mostly studied looking at the superior colliculus.

00:32:29.683 --> 00:32:33.763
So what's the leverage you get looking at that problem? at the cortical level?

00:32:34.343 --> 00:32:39.163
Well, it's because the cortex is necessary for you to report what you,

00:32:39.323 --> 00:32:41.863
to perceive where you've heard the sound from.

00:32:42.063 --> 00:32:46.823
So for example, you can take a cat and you can put it and train it in a sound

00:32:46.823 --> 00:32:50.743
booth to go into the middle of the booth. You can play a sound from one of the speakers.

00:32:50.883 --> 00:32:53.783
It will orient towards the sound and walk towards it and get a treat.

00:32:54.803 --> 00:33:00.443
You lesion the cerebral cortex, A1 and around it, and then the cat doesn't do that.

00:33:00.663 --> 00:33:04.563
So the sound comes from the contralesional space, sometimes it will move its

00:33:04.563 --> 00:33:06.863
pinna, sometimes it will move its head and face the speaker,

00:33:07.003 --> 00:33:08.523
and then it will walk to a random one.

00:33:09.223 --> 00:33:13.183
And what Hugh Hefner interpreted this to me, it happens in monkeys too,

00:33:13.283 --> 00:33:14.143
he's done this in monkeys,

00:33:14.343 --> 00:33:18.743
and what he said is, there is neural circuitry that tells you where the sound

00:33:18.743 --> 00:33:23.663
came from, collicular and brainstem, etc., but you don't have the percept that

00:33:23.663 --> 00:33:24.563
that's where the sound came.

00:33:24.743 --> 00:33:27.803
So when you're going to do something like walk toward where the sound was,

00:33:28.063 --> 00:33:31.083
you really have no conscious percept of where it came from.

00:33:31.543 --> 00:33:35.343
Even if you're looking right at the speaker, because you instinctively turned your head toward it.

00:33:35.663 --> 00:33:42.403
So that means that the target setting is missing, or the transformation of identifying

00:33:42.403 --> 00:33:46.643
the target and location into a goal-oriented action is missing, or both?

00:33:47.043 --> 00:33:49.003
At least the latter. Okay.

00:33:50.185 --> 00:33:55.885
So then, wouldn't you expect the deficit to not so much be affected by areas

00:33:55.885 --> 00:34:00.325
that are so close to auditory processing, but more those that are close to working

00:34:00.325 --> 00:34:03.245
memory and goal setting, frontal areas, and so on?

00:34:03.725 --> 00:34:10.565
Well, probably because we see this kind of behavior in all kinds of mammals.

00:34:10.865 --> 00:34:16.865
You don't have to have a whole big frontal cortex for you to show this deficit, right?

00:34:16.865 --> 00:34:23.645
So Andy King has done a bunch of really elegant studies where he's shown that

00:34:23.645 --> 00:34:29.825
those simple interpretations from the psychophysics from the 70s and 80s doesn't

00:34:29.825 --> 00:34:32.605
really hold in ferrets and probably in any real animal.

00:34:33.005 --> 00:34:37.345
If you have long enough stimuli and allow the animal to move its head,

00:34:37.405 --> 00:34:41.885
as was being discussed, then you can lesion A1 and there's not really much of a deficit.

00:34:42.445 --> 00:34:45.865
You have to lesion more than just A1 for that to not work.

00:34:46.445 --> 00:34:51.265
What's interesting in his studies that I find really remarkable is if you lesion

00:34:51.265 --> 00:34:58.905
A1 in a ferret, and then he can overcome this deficit, but then if you plug one ear,

00:34:59.185 --> 00:35:02.465
a normal ferret will adjust to that and be able to localize fine,

00:35:02.685 --> 00:35:04.305
but not one with an A1 lesion.

00:35:04.465 --> 00:35:08.305
Okay, so A1 is necessary for you to learn how to use these new cues,

00:35:08.465 --> 00:35:13.645
right? And presumably a place like CL in the macaque monkey is where you're

00:35:13.645 --> 00:35:15.785
really doing the sound localization, right?

00:35:16.045 --> 00:35:19.985
And it's not necessarily A1 specific, but it is cortically mediated.

00:35:20.705 --> 00:35:24.405
Because in that sense, in the macaque brain, you've compared quite a number

00:35:24.405 --> 00:35:30.505
of areas to see where would you find the specificity to the location of the sound source.

00:35:32.385 --> 00:35:37.685
And that's actually surprisingly specific. Only one area jumped out that really

00:35:37.685 --> 00:35:42.885
gave you this. the strongest location-modulated response, and actually A1 wasn't

00:35:42.885 --> 00:35:45.385
one of them. So how do you explain that?

00:35:45.765 --> 00:35:52.285
Well, the thinking is that the stimulus comes into A1, A1 contains all of the

00:35:52.285 --> 00:35:53.185
necessary information,

00:35:53.905 --> 00:35:58.765
and then it projects to Cl, and the Cl neurons are specific in the way that

00:35:58.765 --> 00:36:01.945
they interpret that information, and they get the nice small receptive fields,

00:36:02.045 --> 00:36:04.205
which allows then some other area, not Cl,

00:36:04.985 --> 00:36:09.805
to make some sense out of that, and give the monkey the percept of where that

00:36:09.805 --> 00:36:10.845
sound actually came from.

00:36:12.403 --> 00:36:20.283
So, but area CL, if that gives you the percept, is then CL talking to another

00:36:20.283 --> 00:36:23.183
area to translate that to a goal-oriented action?

00:36:23.503 --> 00:36:27.743
Presumably, that's how it works, yeah. Probably the caudal parabelt is going to be critical.

00:36:27.863 --> 00:36:32.043
I imagine that if I did my experiments there, I would see very good correlations as well.

00:36:32.683 --> 00:36:35.543
But then, what's the latency of the response in CL?

00:36:36.803 --> 00:36:41.003
Normally, if you're not old, it's a little bit longer. and so

00:36:41.003 --> 00:36:43.723
Joseph Rauschecker did a really nice study quite some

00:36:43.723 --> 00:36:46.403
time ago and what he found was that if you

00:36:46.403 --> 00:36:54.003
lesion a1 the responses in CM which is its neighbor good to tones goes away

00:36:54.003 --> 00:36:58.283
but to noise stays so the thinking is that there's both noise and tone information

00:36:58.283 --> 00:37:02.963
coming from the thalamus and the cortex to CL and it's getting combined and

00:37:02.963 --> 00:37:05.783
if you take it away the tone information is lost Okay,

00:37:05.863 --> 00:37:11.043
but now the stimuli you use were largely white noise bursts. White noise, yeah.

00:37:11.163 --> 00:37:17.523
Okay, so I could argue then from the perspective of having an auditory percept, that's rather reduced.

00:37:18.083 --> 00:37:23.523
Oh, absolutely. So if we now would start to use, let's say, calls of other monkeys

00:37:23.523 --> 00:37:28.863
or sounds of predators, would you expect that the result might look different,

00:37:28.963 --> 00:37:31.103
that you might find more specificity in other areas?

00:37:32.763 --> 00:37:35.463
More specificity for the different call for example

00:37:35.463 --> 00:37:38.163
no because now we have meaningful sounds actually you use a sound that is

00:37:38.163 --> 00:37:41.063
not meaningful right well it's meaningful in the sense that you had to figure

00:37:41.063 --> 00:37:43.943
out where it was coming from oh you got a reward to it as well you got a reward

00:37:43.943 --> 00:37:49.203
yeah right so i made it meaningful exactly okay okay okay so that that might

00:37:49.203 --> 00:37:53.423
have helped and that's okay you're right exactly so the other the other thing

00:37:53.423 --> 00:37:56.883
you done um so now we have we have of localization,

00:37:57.063 --> 00:38:00.623
we are looking at issues of adult cortical plasticity.

00:38:00.623 --> 00:38:04.163
Now the famous experiments people would do with, let's say, the orienting system

00:38:04.163 --> 00:38:08.123
of the superior colliculus, you put prisms on the owl monkey or,

00:38:08.123 --> 00:38:11.443
or no, on owls, sorry, on barn owls.

00:38:11.503 --> 00:38:16.823
And then, and then you can show that sort of the maps get dramatically reorganized

00:38:16.823 --> 00:38:19.803
also, also in adults, so that the orienting response is aligned.

00:38:21.657 --> 00:38:26.377
Would you expect to see something comparable in in your monkeys when we start

00:38:26.377 --> 00:38:33.857
to sort of uh distort this mapping of sound to location would you also see similar re-mappings in cl,

00:38:34.697 --> 00:38:39.917
oh i predict that you would right as soon as the monkey was able to to do well

00:38:39.917 --> 00:38:44.837
at localizing it my prediction would be that cl would show changes that would

00:38:44.837 --> 00:38:48.117
be really interesting to look at because it's not clear.

00:38:49.117 --> 00:38:53.297
Since there's no topography, you can't really see, oh, these cells changed where

00:38:53.297 --> 00:38:54.837
their best direction was, right?

00:38:54.917 --> 00:38:57.457
You wouldn't be able to see that, but you would see that at the end of the day,

00:38:57.517 --> 00:39:00.897
they all changed appropriately for that to happen.

00:39:01.377 --> 00:39:07.577
Right. Yeah. But to trigger the goal or movement, do we see something similar as we see in vision?

00:39:07.717 --> 00:39:11.297
Like in vision, you might go to frontal eye field and then sort of communicate

00:39:11.297 --> 00:39:14.137
with the colliculus to then set a gaze direction,

00:39:14.137 --> 00:39:16.837
reaction would you believe it's a similar pathway that would

00:39:16.837 --> 00:39:20.537
trigger the orienting response to altruistic stimulus and then the the

00:39:20.537 --> 00:39:23.397
movement to the source i would

00:39:23.397 --> 00:39:26.137
imagine that that would be the case but it might be very much

00:39:26.137 --> 00:39:32.657
different than that why is that well the visual system and gaze shifts are all

00:39:32.657 --> 00:39:38.517
dependent on what you see is depending on where your gaze is right and so in

00:39:38.517 --> 00:39:43.237
the auditory system that's not quite as much so you saw the um psychophysical

00:39:43.237 --> 00:39:44.517
results for these pretty,

00:39:44.577 --> 00:39:46.837
you know, fairly not-quiet sounds,

00:39:47.137 --> 00:39:50.257
it didn't really matter where the sound was coming from, you could tell where it was, right?

00:39:50.317 --> 00:39:52.697
So that would be gaze-independent, right?

00:39:53.017 --> 00:39:57.797
So if you had to orient to the spirit of clickless, you know,

00:39:57.817 --> 00:40:01.437
how exact, it wouldn't surprise me if that was a little bit or a lot different

00:40:01.437 --> 00:40:05.037
than the visual system just for that reason, that the gaze-shifting isn't part

00:40:05.037 --> 00:40:07.997
of the sensory input, right? True.

00:40:08.797 --> 00:40:12.737
But the other thing is also for the auditory stimuli, you might have more ambiguity

00:40:12.737 --> 00:40:15.877
because the visual scene is in front of you and you deal with it,

00:40:15.917 --> 00:40:17.417
and so the information is out there.

00:40:17.777 --> 00:40:24.137
Well, in an auditory display, the source localization might be more difficult

00:40:24.137 --> 00:40:28.577
that you would have to make several orienting movements before you really can pinpoint the source.

00:40:29.877 --> 00:40:35.617
Right. So the amount of error that you will make if you, I didn't show these

00:40:35.617 --> 00:40:39.377
data, but if you're in a dark, you hear a sound, and I ask you to point your

00:40:39.377 --> 00:40:42.477
head toward the sound, the amount of error you make is considerable.

00:40:42.737 --> 00:40:46.297
So that means it might be more an iterative process.

00:40:46.417 --> 00:40:52.097
You do several orienting responses before you really have targeted the source.

00:40:52.417 --> 00:40:57.537
Would that mean that areas like CL have more of a working memory property that

00:40:57.537 --> 00:41:02.457
you might find in more vision-oriented cortical systems?

00:41:02.457 --> 00:41:08.037
That's a good question and the jury is out on that because the auditory stimuli

00:41:08.037 --> 00:41:12.097
are quite a bit different than what you see with visual stimuli.

00:41:15.383 --> 00:41:19.943
They're temporary. You don't have really a lasting auditory stimulus.

00:41:20.463 --> 00:41:24.383
It's pretty rare in nature that something just hums for a long period of time.

00:41:24.643 --> 00:41:28.063
Whereas if you look at the rock, the rock's there and the rock doesn't move

00:41:28.063 --> 00:41:29.543
and you see the rock the whole time.

00:41:29.843 --> 00:41:35.703
As opposed to, there's debate what's an auditory object, does it even exist or is it an event?

00:41:36.463 --> 00:41:39.743
And maybe an event is a better way to think about it. So the way I think about

00:41:39.743 --> 00:41:42.563
it, what's the auditory system in a primate really supposed to do.

00:41:43.143 --> 00:41:46.763
You're minding your own business. You hear a twig snap behind you.

00:41:46.803 --> 00:41:50.643
You can't see what made the twig snap, right? But you hear it and it's a discrete event.

00:41:50.843 --> 00:41:55.343
And so what do you do? You orient and you have to get within plus or minus 15 degrees of it.

00:41:55.463 --> 00:41:58.643
And then you use your visual system to see what snapped the twig.

00:41:58.663 --> 00:42:01.783
Is it something I can eat? Something that's going to eat me or something doesn't matter, right?

00:42:02.123 --> 00:42:06.203
So it's really not that common in

00:42:06.203 --> 00:42:09.623
a primate like a macaque who's diurnal and

00:42:09.623 --> 00:42:12.423
can use its visual system a lot to really worry that much

00:42:12.423 --> 00:42:15.283
about what it's listening to and now

00:42:15.283 --> 00:42:18.103
a monkey might be very different because it's nocturnal right so

00:42:18.103 --> 00:42:22.703
it doesn't see that well and maybe in that species these kinds of things would

00:42:22.703 --> 00:42:30.023
be quite a bit different right right so then as an um application domain uh

00:42:30.023 --> 00:42:34.183
of your understanding of adult plasticity in in auditory auditory processing

00:42:34.183 --> 00:42:37.983
you also start to look at the aging brain yes right and and how

00:42:38.023 --> 00:42:41.083
the aging brain actually started to change its responses to auditory stimuli.

00:42:41.843 --> 00:42:45.483
So what are the most remarkable differences you found there?

00:42:46.503 --> 00:42:51.403
Well, I was really surprised when I did those studies that the activity of the

00:42:51.403 --> 00:42:53.963
brain is so much greater than it is in younger animals.

00:42:54.023 --> 00:42:59.023
Because I came in with the completely false assumption that the older animals

00:42:59.023 --> 00:43:02.503
would have slower brains that would be less active.

00:43:03.266 --> 00:43:07.946
I thought they would probably be more like an anesthetized rat brain as opposed

00:43:07.946 --> 00:43:10.726
to a weight-behaving monkey brain, which is very active.

00:43:11.466 --> 00:43:15.806
Everything makes sense when you're recording from MT in a macaque because the

00:43:15.806 --> 00:43:19.566
motion goes in a certain direction, the cell screams like crazy,

00:43:19.666 --> 00:43:23.546
and you don't have to be a neuroscientist to go, I bet it's doing something

00:43:23.546 --> 00:43:24.986
with the direction of motion, right?

00:43:25.366 --> 00:43:28.866
It's pretty clean. But the old monkeys, they were more active,

00:43:29.066 --> 00:43:30.906
and they were much more sloppy.

00:43:30.906 --> 00:43:33.606
Right and the older monkeys that we

00:43:33.606 --> 00:43:36.406
were studying at the time acted very old they

00:43:36.406 --> 00:43:39.206
would sleep a lot they didn't seem that astute they didn't

00:43:39.206 --> 00:43:42.126
seem like they were really perceiving things very well and i naturally

00:43:42.126 --> 00:43:44.986
equated that with not much brain activity right as opposed to

00:43:44.986 --> 00:43:47.826
the opposite so it was really surprising to see that and

00:43:47.826 --> 00:43:51.586
when we were doing the studies we were uh double

00:43:51.586 --> 00:43:54.766
checking and triple checking and trying to make sure that we're doing everything right

00:43:54.766 --> 00:43:57.506
that we're not messing up with our window discriminators and we're not

00:43:57.506 --> 00:44:00.386
recording a bunch of noise and we're always is looking for

00:44:00.386 --> 00:44:03.946
60 cycle and all these kinds of things to convince ourselves that no

00:44:03.946 --> 00:44:06.926
the neurons are really firing more they're just not firing

00:44:06.926 --> 00:44:10.246
well right so you get this loud sloppy ugly signal

00:44:10.246 --> 00:44:13.206
which i guess the monkeys themselves

00:44:13.206 --> 00:44:16.626
means they have you know weaker percepts

00:44:16.626 --> 00:44:19.326
right and then that's kind of how it all

00:44:19.326 --> 00:44:22.766
ends up to work but is it

00:44:22.766 --> 00:44:25.406
more is it more like say a noisy brain is there

00:44:25.406 --> 00:44:30.626
there's more noise in the brain or is it is more dynamically modulated that

00:44:30.626 --> 00:44:35.906
also if you present a stimulus the response is still specifically tuned in time

00:44:35.906 --> 00:44:43.286
but it's just much much amplified it's more amplified and it's not as specifically tuned in time.

00:44:44.666 --> 00:44:49.026
So i think one of the things you you mentioned there was the possibility that

00:44:49.026 --> 00:44:53.826
that there was perhaps selective loss of inhibitory systems,

00:44:54.026 --> 00:45:00.306
which we know are important for tuning up the dynamics of perceptual systems

00:45:00.306 --> 00:45:02.726
so that you get a clean percept.

00:45:03.006 --> 00:45:08.066
So I think thinking, for instance, of work in RAT-S1,

00:45:08.926 --> 00:45:15.066
Dan Simons, for instance, has talked about the role of interneurons and shutting

00:45:15.066 --> 00:45:20.406
down some of the activity that you will get through the excitatory networks in cortex.

00:45:20.786 --> 00:45:25.866
And then you have a wave of inhibition coming in to make sure that doesn't get it out of control.

00:45:26.106 --> 00:45:32.566
So that it sounds quite consistent with a network which has got some damage,

00:45:32.606 --> 00:45:35.426
but it's not less activity, as you say.

00:45:35.506 --> 00:45:39.566
It's just a lack of appropriate tuning of the inhibition systems.

00:45:39.966 --> 00:45:46.446
Exactly. So I think what's happening in these older animals is their inhibitory network has, um.

00:45:47.774 --> 00:45:51.714
Crashed, essentially, right? It's not doing as well as it. And it's not just in the cortex.

00:45:51.834 --> 00:45:56.254
It's in the cochlear nucleus and the superior olive and inferior colliculus and the geniculate.

00:45:56.334 --> 00:46:00.774
So it's all along the ascending auditory system that there are these problems.

00:46:00.934 --> 00:46:04.494
And by the time it gets to cortex, it hasn't been very well refined, right?

00:46:04.634 --> 00:46:09.994
And so it's a louder, sloppier system, and it just doesn't get any better at the cortical level.

00:46:10.354 --> 00:46:14.934
And has anybody looked for sort of selected loss of inhibitory?

00:46:15.274 --> 00:46:19.014
Not in the cortex. It's been seen in the brainstem and the midbrain.

00:46:19.074 --> 00:46:21.414
Donald Casper's group has shown these kinds of things.

00:46:21.614 --> 00:46:25.134
And there's all kinds of neurochemical changes that happen down there.

00:46:25.254 --> 00:46:29.674
And it hasn't been studied, certainly not in the macaque auditory cortex.

00:46:29.954 --> 00:46:34.254
And the inferior colliculus is the place most people stop on the way up.

00:46:34.354 --> 00:46:38.934
So the cortex is very complicated and it's got layers and all these kinds of things.

00:46:39.054 --> 00:46:43.354
So it hasn't been studied to my knowledge. And I think, as you also said,

00:46:43.474 --> 00:46:48.694
there's an implication of this, which is that if someone's suffering hearing loss,

00:46:48.894 --> 00:46:54.494
the last thing you want to do is actually shout at them, because they're overstimulated anyway.

00:46:54.494 --> 00:46:59.814
Anyway, so slower and lower is the way to go, right?

00:46:59.894 --> 00:47:04.094
And the other key is if you have a hard time temporally processing.

00:47:04.794 --> 00:47:10.334
Speech sounds are temporally very complex and words don't, the way we talk,

00:47:10.354 --> 00:47:13.914
we don't pause between words. We pause at the stop consonants, right?

00:47:14.314 --> 00:47:18.734
So if you're stopping at only the stop consonants and you're running words together,

00:47:19.034 --> 00:47:22.534
right, which we always do, and we can interpret that easily enough,

00:47:22.534 --> 00:47:25.754
it's hard if you can't hear the modulation right yeah

00:47:25.754 --> 00:47:29.054
so the key to talking to an older person is not necessarily talking

00:47:29.054 --> 00:47:32.374
louder that's not going to really work but to pause between

00:47:32.374 --> 00:47:35.214
words right and that way they can get

00:47:35.214 --> 00:47:37.894
each one easily it's the same thing as when you learn

00:47:37.894 --> 00:47:40.514
a language and you're not very good at it and then you go to

00:47:40.514 --> 00:47:43.494
the country and everybody talks really fast right because you're

00:47:43.494 --> 00:47:46.334
you're doing word word word word word right and

00:47:46.334 --> 00:47:49.314
they're pausing at stop consonants so all these words run

00:47:49.314 --> 00:47:52.334
together right right so it's

00:47:52.334 --> 00:47:55.754
it's same things happening in older people but now

00:47:55.754 --> 00:48:02.314
if you look at that these uh the data you present on um the response to these

00:48:02.314 --> 00:48:06.754
older monkey brains versus the younger monkey brains it's actually the latency

00:48:06.754 --> 00:48:09.094
the response latency in the older

00:48:09.094 --> 00:48:14.474
brains seem shorter as well very much correct yes absolutely so so So,

00:48:14.474 --> 00:48:17.494
but also the gain is just up in the system.

00:48:17.714 --> 00:48:21.234
Right. So how about an alternative could be that you say, okay,

00:48:21.294 --> 00:48:26.594
the cochlea, given also this partly mechanical sensor with the hair cells moving,

00:48:26.794 --> 00:48:29.394
that sort of loses sensitivity.

00:48:30.134 --> 00:48:35.614
And what the brain now must do in response is just crank up the gain and everything that follows. Yes.

00:48:36.521 --> 00:48:40.801
So then it's not so much that we lose inhibition due to aging,

00:48:41.041 --> 00:48:45.281
but we're cranking up the gain because the periphery is less sensitive. Would you buy that?

00:48:45.801 --> 00:48:50.481
That certainly would be a reasonable possibility, but I think the evidence certainly

00:48:50.481 --> 00:48:56.341
from the rodent suggests that what's happening is you have less input coming out of the cochlea.

00:48:56.341 --> 00:49:01.361
And so you adapt to that by doing things that effectively decrease inhibition

00:49:01.361 --> 00:49:05.281
as opposed to increase the signal specifically. I think that's where the evidence

00:49:05.281 --> 00:49:06.781
is headed towards at this point.

00:49:07.021 --> 00:49:14.161
And that's based a lot on neurochemical evidence and the changes in the GABA, GABAergic neurons.

00:49:14.261 --> 00:49:18.841
There are two ways to think about this, right? Because you can think about inhibition as gain control.

00:49:19.121 --> 00:49:23.721
Sure. Right? Or you can think about it as sort of sharpening a response that

00:49:23.721 --> 00:49:25.621
you say, look, I have a bunch of frequencies coming in.

00:49:25.741 --> 00:49:31.661
I set up some sort of wind attack all among them. So I sharpen the most salient

00:49:31.661 --> 00:49:32.921
aspect of my input signal.

00:49:33.101 --> 00:49:35.961
Right. I see the two different views on how you can think about the inhibition.

00:49:36.221 --> 00:49:44.961
So do you see it as a sort of a non-specific regulatory adjustment or loss or a more specific one?

00:49:45.981 --> 00:49:49.381
If I had to guess, which you're making me guess. Yeah, of course.

00:49:49.581 --> 00:49:55.041
So my guess is that it's a little bit of both, but it's probably 70-30 and 70%

00:49:55.041 --> 00:49:59.801
is the sharpening part and the 30% is just kind of a global background at the

00:49:59.801 --> 00:50:04.541
single mantra. And that means, in your opinion, this would not be happening at the cortex.

00:50:05.281 --> 00:50:08.781
This would be all subcortical. I think that's what's happening subcortical,

00:50:08.841 --> 00:50:15.181
and I'm really keen to look in the cortex and see what… The cortex is much more

00:50:15.181 --> 00:50:20.041
well-behaved as far as things like palvarbumin being in GABAergic neurons and things like that,

00:50:20.141 --> 00:50:22.701
and the brainstem, it doesn't really do that, right?

00:50:22.961 --> 00:50:26.721
So I would be keen to look in the cortex and see, is there in fact losses of

00:50:26.721 --> 00:50:30.861
GABAergic neurons and which kinds and how would that, how would that work out?

00:50:31.813 --> 00:50:34.553
Because there is a general loss of neurons in the cortex of these animals.

00:50:35.013 --> 00:50:38.973
And another piece of evidence, I think, that speaks to the inhibitory story

00:50:38.973 --> 00:50:45.933
rather than this amplification story is the sort of oscillation that you see.

00:50:47.213 --> 00:50:52.293
Where the firing is high and then it drops and then it comes back high and then drops again.

00:50:52.493 --> 00:50:57.573
And that suggests some alteration in the cortical dynamics dynamics,

00:50:57.693 --> 00:51:02.333
rather than just a raising of the gain generally.

00:51:02.593 --> 00:51:08.313
And that could easily be consistent with loss of inhibitory neurons.

00:51:08.573 --> 00:51:11.953
So for instance, you've got fast-spiking inhibitory neurons,

00:51:12.053 --> 00:51:16.393
which are coming into the cortex quite soon after the initial excitatory burst.

00:51:16.673 --> 00:51:22.953
And people are suggesting that these are there to damp down the response and

00:51:22.953 --> 00:51:26.393
make sure that the neurons that are corresponding are the ones which are most

00:51:26.393 --> 00:51:29.353
finely tuned, which should be very consistent with what you're saying.

00:51:29.633 --> 00:51:35.273
And if that's lost or that's delayed, then you'd imagine dynamics could be upset

00:51:35.273 --> 00:51:38.353
so that you get these oscillations. Do you mind that, Greg?

00:51:38.693 --> 00:51:42.113
No, that works for me. Because then I think you're both wrong.

00:51:43.153 --> 00:51:47.933
Because I think if Tony's interpretation is correct,

00:51:48.213 --> 00:51:51.793
that would also imply a loss of specificity at the cortical level,

00:51:51.873 --> 00:51:55.013
because I'm tuning down my inhibition at the cortical level,

00:51:55.213 --> 00:51:58.593
which means if I have overlaps in my receptive field responses,

00:51:58.933 --> 00:52:01.073
I lose my ability to filter those out.

00:52:01.333 --> 00:52:05.233
And if I look at your data, you don't necessarily look to specificity.

00:52:05.233 --> 00:52:08.053
The specificity isn't necessarily lost in the response.

00:52:08.213 --> 00:52:11.213
It's just really the amplitude that is strongly affected.

00:52:11.993 --> 00:52:15.413
So I think you're both wrong if you agree with Tony. Well, I would say if you

00:52:15.413 --> 00:52:20.493
look in CL, it's pretty clear that the specificity is lost because the spatial

00:52:20.493 --> 00:52:22.453
tuning is about to sink. So I'm sunk now, you're saying.

00:52:24.013 --> 00:52:28.853
I'm sure we're both right. But there's something else that now worries me because,

00:52:28.973 --> 00:52:32.173
okay, here we go. We're getting older, true for all of us.

00:52:34.353 --> 00:52:37.713
Things getting messed up from my cochlea all the way up to my cortex.

00:52:40.513 --> 00:52:43.693
The responses get strongly amplified to get much stronger.

00:52:45.153 --> 00:52:51.453
But in the meantime, you're telling me we have this sort of continuously plastic cortex.

00:52:52.073 --> 00:52:55.853
So if I'm pumping in higher amplitude signals, why is the cortex not adjusting

00:52:55.853 --> 00:52:58.293
to this? It's almost like sewing together two fingers, right?

00:52:58.353 --> 00:53:02.953
I get a stronger drive onto a certain cluster. I have to reorganize my map.

00:53:03.433 --> 00:53:10.253
So are these maps reorganizing and are they reorganizing them in a functionally related way?

00:53:11.313 --> 00:53:15.053
Right. So how do you put these two things together? How come the old monkey

00:53:15.053 --> 00:53:17.653
doesn't have a plastic brain to adapt to all this?

00:53:17.833 --> 00:53:20.573
I don't know the answer to that, but I can guess again.

00:53:20.933 --> 00:53:25.413
And the way I see what happens is you're in your 40s.

00:53:26.106 --> 00:53:29.866
And your hair cells and your spiral ganglion cells, et cetera,

00:53:29.986 --> 00:53:33.226
are starting not to work as well as they used to.

00:53:33.486 --> 00:53:38.146
And so your brainstem and auditory midbrain is beginning to try to adjust to this.

00:53:38.206 --> 00:53:42.006
And it's adjusting this by messing around with the inhibitory system.

00:53:42.546 --> 00:53:46.166
And your cortex is getting a weird signal, and it's being plastic,

00:53:46.246 --> 00:53:48.606
and it's changing. And you have no symptoms.

00:53:49.166 --> 00:53:52.126
Okay, so it works. It's been working for a while, right?

00:53:52.126 --> 00:53:55.166
By the time you're 78, 79 years old,

00:53:55.346 --> 00:54:01.826
for whatever reason, either loss of neurons or loss of the ability to quickly

00:54:01.826 --> 00:54:07.826
adapt or the fact that you are slowing down quite a bit and you're not re-updating

00:54:07.826 --> 00:54:09.566
your maps as often as you are,

00:54:09.646 --> 00:54:13.046
makes it so that that process breaks down.

00:54:13.046 --> 00:54:16.326
And then you start to do plasticity that's hurting you, right?

00:54:16.386 --> 00:54:20.846
So what happens when you're a little bit older and you're having a conversation

00:54:20.846 --> 00:54:26.246
and you keep making up parts of the words that you're missing and then your

00:54:26.246 --> 00:54:31.006
model of the conversation goes south and you ask a really stupid question and

00:54:31.006 --> 00:54:32.266
you get laughed at, right?

00:54:32.266 --> 00:54:35.146
And how is it as you're older and it's a little bit harder to hear,

00:54:35.246 --> 00:54:40.006
so you have to do this, you start to not go into those situations.

00:54:40.346 --> 00:54:44.306
So all of these things that you're normally doing to make sure that your maps

00:54:44.306 --> 00:54:49.706
are okay and you're doing all right, you start to self-deprive these things.

00:54:49.786 --> 00:54:53.386
And what's that going to do? That's going to drive your maps in a bad direction. Right?

00:54:53.646 --> 00:54:57.486
So use it or lose it, you know, is what's been said.

00:54:57.686 --> 00:55:01.566
So if you start to avoid social situations because you can't hear very well,

00:55:01.646 --> 00:55:03.126
you're not going to get better at hearing.

00:55:03.246 --> 00:55:07.866
Sure. Well, that's why I'm training myself to not be bothered with posing a stupid question.

00:55:10.446 --> 00:55:13.506
It's inoculation against my future hearing loss.

00:55:13.786 --> 00:55:20.826
This does also suggest a potential way of developing a program to reduce hearing loss.

00:55:22.190 --> 00:55:25.610
Hearing training that might help us preserve our hearing as we get older.

00:55:26.490 --> 00:55:31.070
Well, that's in some sense what Mike is also doing with his company, Puzzle Science, right?

00:55:31.210 --> 00:55:35.010
Right, so he's doing that. Nina Dronkers in Chicago is doing this with music.

00:55:35.330 --> 00:55:39.850
So if you use motor input and she's looking at the inferior colliculus,

00:55:39.870 --> 00:55:42.390
but there's no reason it's not being translated throughout the system,

00:55:43.110 --> 00:55:51.150
and that will give you the benefit of keeping your maps up to date and maintaining that standard.

00:55:51.150 --> 00:55:54.430
So if I listen to music, that's going to help me maintain a good map.

00:55:54.950 --> 00:55:57.710
It's better than not listening to music, but not as good as playing music.

00:55:57.930 --> 00:55:59.390
It depends on the amplitude.

00:56:00.330 --> 00:56:02.410
The amplitude? At which you play the music.

00:56:04.030 --> 00:56:06.790
If you play it too loud, that's detrimental. You're right. Yeah.

00:56:08.050 --> 00:56:11.850
Well, we can listen to Tony playing this evening. He knows the whole Beatles

00:56:11.850 --> 00:56:13.690
songbook by heart. It won't be too loud. Okay.

00:56:14.550 --> 00:56:19.350
So, but now, so, so this is really great that, that you also start to touch

00:56:19.350 --> 00:56:22.950
upon these more applied to issues and try to get insight in how we actually

00:56:22.950 --> 00:56:23.850
can deal with hearing loss.

00:56:25.770 --> 00:56:27.350
But on the other hand,

00:56:28.978 --> 00:56:33.638
You could also argue that the disinhibition we might see with aging might actually

00:56:33.638 --> 00:56:36.778
also have a positive effect in cognition.

00:56:36.858 --> 00:56:42.538
We could call it wisdom in some sense, right? Because if people get more disinhibited

00:56:42.538 --> 00:56:47.458
or more disconnected from, let's say, their immediate emotional responses to things,

00:56:47.738 --> 00:56:52.398
they can exist in a state of pure cognition or not. Or fantasy.

00:56:52.858 --> 00:56:55.878
Or pure delusion. Yeah, delusion. yeah

00:56:55.878 --> 00:56:58.838
yeah um so now

00:56:58.838 --> 00:57:02.038
you finished up by your speculations about

00:57:02.038 --> 00:57:04.778
the homunculus right and we'd also sort of

00:57:04.778 --> 00:57:08.378
to think a bit about this question okay if we have this adult plasticity we

00:57:08.378 --> 00:57:11.738
have some handle on these mechanisms how would you actually be able to grow

00:57:11.738 --> 00:57:16.698
a new module into a cortex or remove one right so how can we do that how can

00:57:16.698 --> 00:57:24.198
we add let's say a radar map to our own cortex right so um what got me thinking about that

00:57:24.318 --> 00:57:28.138
was the difference in the somatosensory cortex between old world and new world monkeys.

00:57:28.918 --> 00:57:33.118
Where old world monkeys have a clear area 2, which is non-cutaneous,

00:57:33.298 --> 00:57:40.078
and it's between area 1, which is cutaneous, and area 5, which is visual and non-cutaneous as well.

00:57:40.318 --> 00:57:44.118
And new world monkeys don't have that so much, right? So the experiments by

00:57:44.118 --> 00:57:47.418
Jeff Padberg, it really opened my eyes that there was a big difference,

00:57:47.438 --> 00:57:50.578
and it always was curious to me, how would you actually make a new field?

00:57:50.638 --> 00:57:53.558
Because that's what evolution does, we make new fields and humans have

00:57:53.558 --> 00:57:56.558
more than macaques and the common ancestor had some number less

00:57:56.558 --> 00:58:02.378
than both of us right and so um the fact that when you train a monkey to to

00:58:02.378 --> 00:58:07.678
pay attention or to discriminate a tactile stimulus on their finger and you

00:58:07.678 --> 00:58:12.638
essentially change the morphology and the functional functionality of an entire

00:58:12.638 --> 00:58:15.158
cortical area over that representation area 3a.

00:58:16.398 --> 00:58:20.598
Leads me to think that well that can happen pretty quickly right and so what

00:58:20.598 --> 00:58:22.438
would it take to get a new cortical field.

00:58:23.138 --> 00:58:28.778
If you wanted a new cortical field, you can either change it in a lifetime,

00:58:28.958 --> 00:58:32.918
in weeks, but you lose whatever it used to do because your brain doesn't get

00:58:32.918 --> 00:58:34.998
bigger because it's stuck in the skull, right?

00:58:35.098 --> 00:58:39.518
So one way that you could do this is you could alter your functional cortical

00:58:39.518 --> 00:58:43.018
topography by changing your niche, right?

00:58:43.138 --> 00:58:45.998
And then when you get the opportunity to get a bigger brain,

00:58:46.198 --> 00:58:50.878
you know, you have that and then that same kind of process can fill in that space, right?

00:58:50.938 --> 00:58:54.998
So it just seems to me that you don't have to wait 30 million years to hope

00:58:54.998 --> 00:58:58.738
that your progenitor cells last a little bit longer so that you get a bigger

00:58:58.738 --> 00:59:01.018
sheet. Then you decide, what am I going to do with it, right?

00:59:01.098 --> 00:59:07.238
You can jumpstart that and already have in place something that's new, right?

00:59:07.438 --> 00:59:10.778
And then when you do get more tissue, you're ready to fill it in with that.

00:59:11.638 --> 00:59:15.978
So that just made sense to me. Okay. But now, isn't the notion of homunculus

00:59:15.978 --> 00:59:18.118
not overrated to start with?

00:59:19.229 --> 00:59:24.329
Like, for instance, it's very literal, almost like a copy of the body in the cortex.

00:59:24.589 --> 00:59:27.349
Also, you already showed this incredible individual variability,

00:59:27.669 --> 00:59:30.649
right, in how you organize the somatotopic maps.

00:59:31.829 --> 00:59:37.229
Moreover, also within the somatotopic region, you might have,

00:59:37.289 --> 00:59:40.689
let's say, duplications of such a map. You might have sub-maps within maps.

00:59:40.809 --> 00:59:45.049
How should we think really about this? How accurate is something like a homunculus?

00:59:45.429 --> 00:59:49.129
Well, it depends on how you measure it. So if you're Penfield and Rasmussen

00:59:49.129 --> 00:59:54.509
and you put your stimulator every 7 to 10 or 12 millimeters across,

00:59:54.669 --> 00:59:56.229
you get a homunculus. Okay.

00:59:56.669 --> 01:00:02.689
Yeah. If you take a microelectrode and you penetrate every 50 to 75 microns,

01:00:02.849 --> 01:00:06.189
it's not super duper clean, right?

01:00:06.249 --> 01:00:11.289
The map of the hand in the owl monkey is, except that the hairy skin is in all

01:00:11.289 --> 01:00:15.289
kinds of different weird places and there's overlap and it looks pretty good.

01:00:17.209 --> 01:00:20.889
Um, other parts of the body are not so much, right? So when the receptive fields

01:00:20.889 --> 01:00:25.589
start to get big, like on the rump or the leg, it starts to get a little bit sloppy too. Right.

01:00:27.529 --> 01:00:35.429
So, um, now to, to finish, finish up the, our, our debate here or, uh, our conversation,

01:00:35.689 --> 01:00:41.109
um, so, so you started out to look at this issue of adult plasticity where at

01:00:41.109 --> 01:00:44.649
the beginning, so actually getting this back on the map again, right, which is great.

01:00:45.749 --> 01:00:49.549
And to travel all the way through, let's say, some of the sensory systems,

01:00:49.749 --> 01:00:51.969
motor systems, auditory systems.

01:00:52.249 --> 01:00:57.029
So now if I really want to follow in a tradition in which you have me working,

01:00:57.309 --> 01:01:01.529
what is Greg's law that we should adhere to understand the brain?

01:01:01.709 --> 01:01:03.349
Greg's law to understand the brain.

01:01:04.629 --> 01:01:06.069
It's harder than it looks.

01:01:07.649 --> 01:01:08.669
Harder than it seems?

01:01:10.623 --> 01:01:13.383
Yeah, well, I think if you want to understand the cerebral cortex,

01:01:13.423 --> 01:01:18.403
and this is what I've done, so that's why I think it, you should not get stuck in a cortical area.

01:01:19.023 --> 01:01:25.323
You should not say, I want to understand how the cerebral cortex operates to

01:01:25.323 --> 01:01:29.903
provide perceptions, and to do that, I'm going to spend 30 years studying MT.

01:01:29.903 --> 01:01:33.063
Okay so i think you're not going to gain the insights that

01:01:33.063 --> 01:01:35.883
you need unless you look across a bunch of different

01:01:35.883 --> 01:01:39.123
cortical areas and you know a bunch

01:01:39.123 --> 01:01:42.823
of different species would probably be pretty smart too but i'm i'm not only

01:01:42.823 --> 01:01:46.783
a cortical snob i'm a primate snob too so i've been i've been sticking to the

01:01:46.783 --> 01:01:52.043
primate but um one nice thing about being a close colleague of leah kribitzer's

01:01:52.043 --> 01:01:56.783
is that she studies all kinds of different animals and so i i know all about these things,

01:01:56.903 --> 01:02:00.863
and it's something that you can really use to gain your insight.

01:02:01.023 --> 01:02:04.423
So if you're not going to study them yourself, you should at least bone up on

01:02:04.423 --> 01:02:08.943
the literature and talk to somebody who does so they can keep you grounded.

01:02:09.203 --> 01:02:12.603
So embrace variability would be it. That would be a good one, yeah.

01:02:12.803 --> 01:02:16.763
Okay, so five years from now, Tony likes traveling. He comes visit you in Davis.

01:02:17.023 --> 01:02:22.003
And he's going to test, check whether you verify the hypothesis that you're

01:02:22.003 --> 01:02:22.803
going to share with us now.

01:02:22.803 --> 01:02:25.883
So what's the most important hypothesis

01:02:25.883 --> 01:02:29.343
you want to see verified in the time frame of five years in

01:02:29.343 --> 01:02:32.203
a time frame of five years so that's uh in in

01:02:32.203 --> 01:02:38.163
the monkey world that's not very much time whatsoever right so um what i would

01:02:38.163 --> 01:02:43.823
like to see is um how is it that this proposed parallel pathway with the rostral

01:02:43.823 --> 01:02:50.143
and the the caudal what uh where in the rostral what pathway what is that really like?

01:02:50.323 --> 01:02:53.723
I mean, how is it really doing? My money is that it's not going to be quite

01:02:53.723 --> 01:02:58.123
as clean as it is in the visual system, but it's going to probably provide us

01:02:58.123 --> 01:03:00.043
a lot more insights, I hope anyway.

01:03:00.623 --> 01:03:04.683
Into how not just the visual system, but also the mass sensory system and others,

01:03:04.783 --> 01:03:09.363
how you actually do the binding of these objects into, you know, a single kind of thing.

01:03:09.483 --> 01:03:14.283
So I'm hopeful that five years from now when Tony comes by, I'll be able to

01:03:14.283 --> 01:03:16.503
tell him almost how it works.

01:03:17.103 --> 01:03:20.783
Great. Rick and Zoan, thank you very much for this conversation. Thank you. Thank you.

01:03:36.748 --> 01:03:41.808
For more interviews, recorded lectures, or upcoming conferences in the field

01:03:41.808 --> 01:03:48.048
of biometrics and biohybrid systems, go to csnnetwork.eu.

01:03:48.708 --> 01:03:50.228
And thank you for listening.

01:03:55.308 --> 01:03:59.768
You're off to? All right. Well, thank you, guys. That was, never done that before.

01:04:00.388 --> 01:04:03.728
I thought it was cool. Yeah, it's good. I like it. Yeah. Nice.

01:04:03.728 --> 01:04:04.828
Thank you. You're welcome.

01:04:06.028 --> 01:04:08.948
It's great. I didn't know you were instructed by Leia to talk about this.

01:04:09.508 --> 01:04:12.488
Yeah. If Leia wouldn't have given you any instruction, what would you have talked about?

01:04:13.488 --> 01:04:16.428
Probably I would have talked all about the auditory cortex. Okay.

01:04:16.828 --> 01:04:20.368
That's all I would have talked about. Right. And I would have done the,

01:04:20.428 --> 01:04:24.468
we also looked at a whole bunch of different things in the temporal domain.

01:04:24.728 --> 01:04:27.408
Uh-huh. Right? So I would have talked about that. Ah, cool. Okay.

01:04:27.468 --> 01:04:30.748
I didn't know that. Yeah. Yeah. But all cortical. All cortical. Okay.

01:04:31.828 --> 01:04:39.028
What's the time window which cortex can reliably represent interval? information on its own.

01:04:41.228 --> 01:04:45.668
Like how fast can it? No, no. If I give you a beep, let's say,

01:04:45.708 --> 01:04:49.888
or you have to hold a stimulus in memory for a certain amount of time,

01:04:50.048 --> 01:04:54.368
what's the time window which Cortex can do this reliably in a task?

01:04:54.968 --> 01:04:59.128
Uh, the time window in which a monkey can reliably do a, say,

01:04:59.148 --> 01:05:03.508
a delayed match to non-sample or something like that? Seconds. Seconds.

01:05:04.328 --> 01:05:07.608
And where is this information stored? Is it refrigerating in the cortical circuit?

01:05:08.448 --> 01:05:12.268
Yeah, that's a question you'd have to ask one of five or six of Morton,

01:05:12.288 --> 01:05:14.388
Michigan's postdocs. They couldn't figure it out.

01:05:15.428 --> 01:05:18.608
But do you have any idea? Is it the cerebellum or the cortex?

01:05:18.848 --> 01:05:21.528
It would probably have to be, I think it would probably be in the cerebellum.

01:05:21.828 --> 01:05:26.708
So it gets to the, the auditory system gets to the thing that I said before.

01:05:27.368 --> 01:05:31.548
I can hold this in my visual working memory really well, right?

01:05:31.548 --> 01:05:35.548
But the auditory system doesn't have objects like this that you have to do.

01:05:35.748 --> 01:05:41.028
And so it's been a real difficult thing for the field to figure out what you

01:05:41.028 --> 01:05:43.568
can reliably remember in memory.

01:05:43.828 --> 01:05:46.908
So you can train a monkey to listen to a bunch of sounds, and what it hears

01:05:46.908 --> 01:05:49.868
is boop, boop, beep, to let go. And it can remember that all day,

01:05:49.968 --> 01:05:51.528
and that'll be held in there forever.

01:05:52.168 --> 01:05:55.168
But to do the classic, beep, beep, boop.

01:05:56.588 --> 01:05:59.628
Okay, now is it beep, boop, beep, or beep, beep, boop? And it's like,

01:05:59.748 --> 01:06:02.068
I can't remember, you know?

01:06:02.188 --> 01:06:08.728
Right. And so all of those, Mort tried for years to do those ablation behavior

01:06:08.728 --> 01:06:14.448
experiments that made him famous in the auditory domain and just couldn't get the stupid monkeys to,

01:06:14.568 --> 01:06:18.768
I mean, he started those when I was in Mort's lab. Right. Simply Guitar?