WEBVTT

00:00:03.537 --> 00:00:10.497
This is the Convergent Science Network podcast. Leading researchers in the domain

00:00:10.497 --> 00:00:16.777
of neuroscience, brain theory and technology are interviewed by Paul Verschoor and Tony Prescott.

00:00:25.957 --> 00:00:29.997
This is Paul Verschoor with the Convergent Science Network, work together with

00:00:29.997 --> 00:00:32.957
Tony Prescott and our guest Alex Kacelnik.

00:00:33.137 --> 00:00:38.997
And Alex is a biologist who has been studying in great detail the extraordinary,

00:00:39.277 --> 00:00:43.357
also cognitive capabilities of a range of animals, in particular birds.

00:00:44.377 --> 00:00:50.317
So Alex, why do you think birds are a good target species to study to understand animal cognition?

00:00:51.277 --> 00:00:56.257
Well, as in many of these things, some of these things have history more than

00:00:56.257 --> 00:01:01.537
real logical explanation. I mean, one starts in a system and keeps going.

00:01:01.997 --> 00:01:09.237
But in reality, birds have certain good properties from the point of view of studying behavior,

00:01:09.777 --> 00:01:15.357
in the style that I'm interested in, because A, they are diurnal,

00:01:15.357 --> 00:01:19.577
and so you can study them in daytime, and they are very visible.

00:01:19.737 --> 00:01:23.357
So if you want to study them in the field, you can see what they are doing,

00:01:23.397 --> 00:01:26.917
as opposed opposed to seeing rats, which you have to go down into the sewage

00:01:26.917 --> 00:01:29.717
and studying them at dark.

00:01:30.097 --> 00:01:33.657
So it's much more complex.

00:01:33.917 --> 00:01:41.097
So birds have a wealth of stereotype behavior, which has occasionally been misinterpreted,

00:01:41.117 --> 00:01:46.857
very often been misinterpreted as implying lack of the capability for flexible behavior.

00:01:47.917 --> 00:01:51.737
And when, in fact, having the ability to do certain things without having learned

00:01:51.737 --> 00:01:59.657
them doesn't mean that you cannot combine them with innovation and new discoveries

00:01:59.657 --> 00:02:01.997
and cognitively demanding tasks.

00:02:02.237 --> 00:02:10.917
So I started, I guess, well, if I ignore my very early years,

00:02:11.757 --> 00:02:16.157
I started working on birds because I could...

00:02:17.296 --> 00:02:23.716
Look at them both in the field and in the lab, I could create conditions for

00:02:23.716 --> 00:02:25.896
relatively simple interfaces.

00:02:26.136 --> 00:02:28.196
They don't use, although they

00:02:28.196 --> 00:02:31.256
use more than ghosts believe, they don't use smell as much as mammals.

00:02:31.456 --> 00:02:35.736
It doesn't mean that they don't use olfactory cues, but they are less olfactory

00:02:35.736 --> 00:02:40.956
than mammals, meaning that visual stimuli, which we can easily understand and

00:02:40.956 --> 00:02:45.436
program, can kind of easily be implemented in the laboratory.

00:02:47.476 --> 00:02:53.916
And they have a life cycle, regular life cycle in a way.

00:02:54.096 --> 00:03:01.156
But now you also placed that study of cognition in a very clear evolutionary perspective, right?

00:03:01.196 --> 00:03:06.216
So you really see that cognition evolved to actually address a very specific

00:03:06.216 --> 00:03:11.796
set of features of the interaction with the world, which is that the kinds of

00:03:11.796 --> 00:03:14.876
problems we encounter to survive might come in different qualities.

00:03:14.876 --> 00:03:17.156
So, what's the distinction that you see there?

00:03:17.376 --> 00:03:21.796
Yeah, you're absolutely right. I see myself fundamentally as a biologist,

00:03:21.896 --> 00:03:29.616
and I see cognition as yet another part of the toolkit of the animal to adjust to its environment.

00:03:29.856 --> 00:03:36.436
In a sense, it's not different from the way Herbert Simon conceived the understanding of the mind.

00:03:36.436 --> 00:03:42.576
I mean, you have, you know, the famous metaphor of the scissors where you have

00:03:42.576 --> 00:03:45.556
the two blades to make the scissor cut.

00:03:45.796 --> 00:03:50.476
And then if you think of understanding the mind without understanding the environment

00:03:50.476 --> 00:03:54.896
in which it evolves, not knowing which problems the mind has evolved to solve,

00:03:54.976 --> 00:03:57.736
it's going to be a handicap for you to understand the mind.

00:03:57.736 --> 00:04:05.016
The mind, and I use the term very loosely here, but they say the product of cognition in general,

00:04:05.176 --> 00:04:14.516
has not evolved to do the fanciest possible calculations or to just give joy

00:04:14.516 --> 00:04:19.936
to either observers or the owners of those minds.

00:04:20.056 --> 00:04:24.616
It has evolved because it's a good way to solve problems, and the problems are

00:04:24.616 --> 00:04:31.936
determined by natural selection. So you try to link both and evolve towards

00:04:31.936 --> 00:04:33.536
a better understanding.

00:04:33.816 --> 00:04:36.296
Right. But it calls me the case that in evolution, let's say.

00:04:37.011 --> 00:04:42.551
You can get away with pre-wiring stereotype behaviors for problems that at an

00:04:42.551 --> 00:04:45.091
evolutionary time scale are constant.

00:04:45.471 --> 00:04:50.111
Let's say to right yourself against gravity might be something you could just

00:04:50.111 --> 00:04:52.611
pre-wire because gravity doesn't change so rapidly,

00:04:52.871 --> 00:04:57.571
while finding the fridge to get a beer might be a problem for which you might

00:04:57.571 --> 00:05:01.251
want to have at least to some extent cognitive capabilities, right?

00:05:01.491 --> 00:05:04.511
Yes, you're absolutely right. In the In the area of decision-making,

00:05:04.691 --> 00:05:10.091
there is a very crucial distinction between what is normally understood as risk

00:05:10.091 --> 00:05:13.691
and what is understood as uncertainty or 90 and uncertainty.

00:05:14.231 --> 00:05:17.331
In the case of risk, you know the probabilities of events.

00:05:17.471 --> 00:05:21.711
And although the world is inherently stochastic, that is, you cannot predict

00:05:21.711 --> 00:05:26.891
exactly the consequences of your actions, you know the probability distributions around yourself.

00:05:27.311 --> 00:05:31.311
In the case of uncertainty, you may not even know the nature of the problem

00:05:31.311 --> 00:05:35.211
very well. You don't know the probabilities of each part of the problem.

00:05:36.111 --> 00:05:42.171
And in evolutionary sense, things that for the individual are a case of uncertainty,

00:05:42.451 --> 00:05:44.831
for evolution may be a case of risk.

00:05:45.471 --> 00:05:49.931
And that in one lifespan, you may not know the probability that,

00:05:49.931 --> 00:05:55.251
say, it's going to be warm in May if you are born in February, for example.

00:05:55.491 --> 00:05:59.571
But evolutionarily speaking, you can know that because that's encoded in your

00:05:59.571 --> 00:06:01.811
genome gnome through the success of your ancestry.

00:06:03.411 --> 00:06:06.391
But also the first model animal that you...

00:06:06.953 --> 00:06:10.893
Presented to us was the siskins, right?

00:06:10.973 --> 00:06:17.673
So you were looking at this problem of mate selection in siskins and you saw

00:06:17.673 --> 00:06:22.793
this as an example of also how ambiguous it is to think about,

00:06:23.433 --> 00:06:27.153
cognition as necessarily helping you in survival, right?

00:06:27.233 --> 00:06:30.293
So what was the message of these experiments with these siskins?

00:06:30.593 --> 00:06:34.953
Well, I was referring to experiments done by the group of Senar,

00:06:35.013 --> 00:06:36.413
Juan Carlos Senar, here in Barcelona.

00:06:36.833 --> 00:06:43.173
And what they did was to study these little birds in which some aspects of female

00:06:43.173 --> 00:06:45.513
mate choice are very well understood.

00:06:45.713 --> 00:06:52.173
The females prefer males which have an enhanced yellow bar in their wings.

00:06:52.473 --> 00:06:56.573
And what they did was to ask the question, why do the females prefer this?

00:06:56.793 --> 00:07:00.733
And they tested the animals in the laboratory and they found that And there's

00:07:00.733 --> 00:07:03.733
a correlation between their ability to solve certain problems,

00:07:03.953 --> 00:07:10.253
how fast they solve it, and the feature that the females are using to select between them.

00:07:10.333 --> 00:07:16.773
So as if females are using the yellow bar as a proxy for picking up a male that's

00:07:16.773 --> 00:07:20.853
going to be good at feeding their young, for example, or at providing their

00:07:20.853 --> 00:07:24.673
young with good genes for being efficient foragers.

00:07:24.673 --> 00:07:31.373
I simply was using their work as an example of the complexities of saying that the male.

00:07:32.273 --> 00:07:37.673
Intelligence may evolve through sexual selection the female role in this but

00:07:37.673 --> 00:07:44.613
the females is actually using a characteristic that we don't fully understand to,

00:07:45.953 --> 00:07:49.393
Develop their preferences, right? So but then I.

00:07:51.541 --> 00:07:59.201
On the other hand, that raises the question, why are not all these syskins,

00:07:59.361 --> 00:08:04.261
let's say, developing a big yellow bar even though they're stupid?

00:08:05.081 --> 00:08:09.601
Because you would easily try to, in that sense, fool. You could fool your potential

00:08:09.601 --> 00:08:11.901
mates. Why is that not happening with these syskins?

00:08:12.241 --> 00:08:16.181
Well, that's taking us, your question takes us to a basic issue of the evolution of communication.

00:08:16.721 --> 00:08:21.741
In communication, you have an emitter and a receiver. Some signal has evolved

00:08:21.741 --> 00:08:27.381
in the emitter because it's a good way to alter the behavior of the receiver

00:08:27.381 --> 00:08:29.081
through the receiver's senses.

00:08:30.021 --> 00:08:35.081
And the question is, it's no question I would do whatever favors me.

00:08:35.161 --> 00:08:43.041
But, of course, the receiver evolves responding to signals in a way that actually benefits it.

00:08:43.241 --> 00:08:48.481
And so why would a receiver accept the signal if there's a frequency-dependent

00:08:48.481 --> 00:08:49.621
issue immediately emerging?

00:08:49.621 --> 00:08:59.221
If too many stupid males had the sign for intelligence, then the preference

00:08:59.221 --> 00:09:03.281
for that clue would disappear and the females would be selected out.

00:09:03.341 --> 00:09:04.161
It wouldn't convey information.

00:09:04.681 --> 00:09:12.321
So you have a constant balance between the advantage of the psychology of the

00:09:12.321 --> 00:09:14.701
receiver and the benefit to the emitter.

00:09:14.701 --> 00:09:19.261
So communication is not about accuracy of conveying information.

00:09:19.621 --> 00:09:21.201
It's about efficiency of manipulation.

00:09:21.661 --> 00:09:30.121
And for the receiver, it's efficiency of decoding variables in the emitter that are of use to self.

00:09:30.381 --> 00:09:35.401
And I was just trying to illustrate a little bit of that. Right, exactly.

00:09:35.681 --> 00:09:40.941
So then the core of your experiments that we're going to talk about later with

00:09:40.941 --> 00:09:44.901
a number of birds, bird species…,

00:09:46.527 --> 00:09:48.967
It all turns around this issue of problem-solving and insight.

00:09:49.347 --> 00:09:53.787
And you started with this very classic example of colors, chimpanzee experiments.

00:09:55.587 --> 00:10:01.207
So why is that such a critical experiment or such a strong illustration of the

00:10:01.207 --> 00:10:02.807
kind of paradigm that you're interested in?

00:10:04.187 --> 00:10:10.107
Just to remember, these experiments were done in the early 20th century.

00:10:10.107 --> 00:10:16.807
And the basis was to present the chimps with a novel problem and see how they

00:10:16.807 --> 00:10:20.927
innovated with instruments around themselves to achieve a solution.

00:10:21.167 --> 00:10:27.787
At the time, that was interpreted as evidence for the existence of insight and

00:10:27.787 --> 00:10:33.227
of mental modelling and trial and error in the mind by the chimps.

00:10:33.227 --> 00:10:40.887
And I use this as an example of the weakness of our tendency to project the

00:10:40.887 --> 00:10:45.167
way we think we solve problems into the data we collect from other species.

00:10:46.047 --> 00:10:53.847
I use as an example of this some work from some of the people of the American Behaviorist School,

00:10:54.007 --> 00:10:58.367
the Scandinavian School, particularly Robert Epstein, who showed that if you

00:10:58.367 --> 00:11:04.327
train pigeons on the components of a task, and one day you offer a problem that they have never faced,

00:11:04.527 --> 00:11:11.387
but where the different components can be chained together to achieve a solution, the pigeons also did it.

00:11:11.547 --> 00:11:15.227
And of course, this had not been done systematically in the chimps' experiments,

00:11:15.467 --> 00:11:18.987
but the chimps have their personal experience, which they could do it.

00:11:19.207 --> 00:11:24.167
And the issue I raised was, if the pigeon does the same as the chimp,

00:11:24.307 --> 00:11:29.407
should we conclude that none of them is intelligent or that both of them are?

00:11:29.587 --> 00:11:34.247
And how do we separate them? And that was just a springboard to start designing

00:11:34.247 --> 00:11:39.447
cases where we see animals solving problems,

00:11:39.707 --> 00:11:43.967
which are novel in the way they are presented, but somehow resonate with their

00:11:43.967 --> 00:11:44.807
experience, obviously.

00:11:45.507 --> 00:11:50.687
And how could then unravel, really, what is the task for the organism?

00:11:50.927 --> 00:11:56.287
How is it doing? And this is where we connect with people working with relatively autonomous robots.

00:11:56.667 --> 00:12:01.547
But now, how I take your Epstein example is that,

00:12:02.167 --> 00:12:06.547
so Epstein had his four equations of what he called generative behavior,

00:12:06.547 --> 00:12:11.827
which are very reminiscent, obviously, of also Thorndike's law of effect.

00:12:11.947 --> 00:12:15.447
And that essentially means that as a learning organism, you're really at the

00:12:15.447 --> 00:12:16.467
mercy of your environment.

00:12:16.567 --> 00:12:19.267
Your environment is instructing you about what you have to do.

00:12:19.347 --> 00:12:24.327
And to build a chain, you do have to receive specific reinforcement that tells

00:12:24.327 --> 00:12:27.947
you, well, A and B belong together and should be executed together. Right.

00:12:28.388 --> 00:12:33.848
So that's sort of this empty organism notion.

00:12:34.768 --> 00:12:39.128
But now, in your case, if you look at insight, insight might require something

00:12:39.128 --> 00:12:41.488
like an internal model, internal representation.

00:12:42.568 --> 00:12:45.828
So where do you place your own thinking in between those extremes?

00:12:46.108 --> 00:12:51.628
Do you see this as relying on internal models, or do you really think that this

00:12:51.628 --> 00:12:53.068
kind of complex behavior,

00:12:53.428 --> 00:12:58.068
problem-solving behavior that might look like insight can really be instructed

00:12:58.068 --> 00:13:02.428
by direct reinforcements from the environment as the behaviorists would have it.

00:13:02.748 --> 00:13:07.208
Well, there are several aspects to your very complex question.

00:13:07.548 --> 00:13:12.568
One of them is that a totally instructed organism doesn't really exist.

00:13:12.828 --> 00:13:17.248
I mean, as you well know, we filter the information that the world is giving us.

00:13:17.288 --> 00:13:21.308
We selectively take particular relations in the world.

00:13:21.388 --> 00:13:24.788
I mean, we have a selective perception, selective processes of conceptualization.

00:13:24.788 --> 00:13:29.308
And different species differ in what they filter of their environment and how

00:13:29.308 --> 00:13:32.328
different in their gaze and then how they look at the world.

00:13:32.648 --> 00:13:37.008
And as a biologist, I'm very keen on taking that into account.

00:13:37.188 --> 00:13:40.928
What is it that the organism is actually trying to extract as significant information?

00:13:41.508 --> 00:13:45.308
On the other hand, the job that learning is doing for the animal,

00:13:45.428 --> 00:13:48.868
if we go back to the notions I mentioned before of risk and uncertainty.

00:13:49.528 --> 00:13:52.808
What learning is doing is transforming uncertainty into risk.

00:13:52.808 --> 00:13:58.288
So you start an animal with not knowing the problem it's in,

00:13:58.408 --> 00:14:02.328
other than with what natural selection has actually encoded it for,

00:14:02.488 --> 00:14:09.148
and evolving by its own experience into actually plugging in the parameters

00:14:09.148 --> 00:14:12.188
of its real life circumstances.

00:14:12.188 --> 00:14:19.008
I mean, what world I am in, and what are the affordances of the objects around

00:14:19.008 --> 00:14:24.328
me, and what are the laws of this, even the social environment in which I'm moving.

00:14:24.668 --> 00:14:29.108
And that cannot be anticipated by natural selection, so that job is done by learning.

00:14:29.888 --> 00:14:32.348
Now, you asked me about insight. Right.

00:14:33.367 --> 00:14:40.227
I don't want to really define myself as, it's not the kind of thing you can be for or against.

00:14:40.447 --> 00:14:46.927
I mean, in a lot of cases, it is very frequent to over-interpret data as demonstrating

00:14:46.927 --> 00:14:50.067
that this could not be done other than by insight.

00:14:50.487 --> 00:14:56.107
But in many cases, insight is just labeling something like the animal could

00:14:56.107 --> 00:14:57.807
not do it, and then it did it.

00:14:57.807 --> 00:15:01.687
And this discontinuity in

00:15:01.687 --> 00:15:05.007
the data is explained by something miraculous that happened

00:15:05.007 --> 00:15:08.947
in the animal that we don't understand better by putting a name to it so if

00:15:08.947 --> 00:15:14.007
we could actually show what mental operations the animal is doing and have some

00:15:14.007 --> 00:15:19.647
handle experimentally or quantitatively in some kind of way then i i'm happy

00:15:19.647 --> 00:15:24.187
with mentalistic interpretations but they have to have a handle Right, exactly.

00:15:53.587 --> 00:15:56.547
Of an image then it takes longer to actually

00:15:56.547 --> 00:16:04.287
do it the greater the angle and if you ask individuals why is how they do it

00:16:04.287 --> 00:16:09.487
they tell you that in their minds they are rotating the the object until they

00:16:09.487 --> 00:16:13.707
see it vertically and then they can actually take a decision so So,

00:16:13.707 --> 00:16:17.567
Delius did the same experiment,

00:16:18.367 --> 00:16:25.467
but found that the pigeons had no extra time for objects that were rotated with

00:16:25.467 --> 00:16:28.087
respect to the training orientation.

00:16:30.347 --> 00:16:35.687
And if you did find it, you could conclude and accept that maybe they were using

00:16:35.687 --> 00:16:38.947
some kind of mental rotation and do further experiments about it.

00:16:39.187 --> 00:16:40.687
But the fact that you didn't,

00:16:41.721 --> 00:16:45.181
doesn't tell you that they don't have the capability for mental rotation.

00:16:45.281 --> 00:16:51.101
It tells you that an animal that flies looking at a horizontal world from above

00:16:51.101 --> 00:16:56.381
and looking from a different perspective doesn't have a priority axis like the

00:16:56.381 --> 00:17:01.761
vertical one and can actually quickly identify images by their pattern regardless

00:17:01.761 --> 00:17:02.961
of whether they are rotated.

00:17:03.961 --> 00:17:07.321
But we can't actually tell what's going on in the mind of the animal.

00:17:08.481 --> 00:17:11.061
But is your statement there also more that you're saying, look,

00:17:11.161 --> 00:17:14.521
you want to link to Epstein and this whole tradition behind him,

00:17:14.581 --> 00:17:18.321
going back to Thorndike and Pavlov, not so much to say, look,

00:17:18.481 --> 00:17:22.201
conceptually, I believe in an empty organism, but in our explanations,

00:17:22.401 --> 00:17:26.181
we should keep it as minimal as possible and really ground it in the data we

00:17:26.181 --> 00:17:29.661
have and not in our anthropomorphic interpretations of this behavior of the

00:17:29.661 --> 00:17:31.761
animals. Yes, absolutely.

00:17:32.141 --> 00:17:37.641
I think from that point of view, I would place myself in the range between the

00:17:37.641 --> 00:17:44.821
mystical psychologist and the killjoy behaviorist.

00:17:44.861 --> 00:17:50.661
I would place myself closer to the killjoy than to the mystical.

00:17:50.661 --> 00:17:53.821
But i admit that a

00:17:53.821 --> 00:17:56.721
lot of a behavior we ourselves observes and collect

00:17:56.721 --> 00:18:00.181
on our animals we don't have an algorithmic model to

00:18:00.181 --> 00:18:03.001
say how does experience translate into these problems also

00:18:03.001 --> 00:18:05.781
and we are looking at these problems to see if we can

00:18:05.781 --> 00:18:10.241
build them yes that was more or less my question but you're saying that the

00:18:10.241 --> 00:18:14.181
key principle here is is parsimony so we look for the simplest explanation of

00:18:14.181 --> 00:18:19.421
behavior and a second principle i guess is continuity that we look for things

00:18:19.421 --> 00:18:24.261
that we see in other animals that might explain this when we see it in birds.

00:18:25.021 --> 00:18:31.041
Is there not some reason to expect maybe that some of these bird species do

00:18:31.041 --> 00:18:34.341
have competences that might not exist in many other animals?

00:18:34.621 --> 00:18:40.401
Should we not be a bit more generous in expecting birds, especially ones with

00:18:40.401 --> 00:18:45.281
larger brains, to have some of the abilities that you might only see, say, in primates?

00:18:46.641 --> 00:18:52.121
Yes, I think that's what you say is empirically validated already.

00:18:52.421 --> 00:19:02.621
I mean, we know already that many problems that others and ourselves have shown

00:19:02.621 --> 00:19:09.461
birds to be capable of solving have not yet been shown as being solved by mammalian

00:19:09.461 --> 00:19:11.381
species, including primates.

00:19:11.381 --> 00:19:17.441
There's no question that every species has particular competencies.

00:19:17.641 --> 00:19:24.041
And if we create a scale where the yardstick is human ways of doing things,

00:19:24.181 --> 00:19:28.681
then the closer you are to humans, the better you're going to score in that

00:19:28.681 --> 00:19:30.061
scale. But that's circular.

00:19:31.541 --> 00:19:37.481
But their target animal in many of your experiments has been the Caledonian crow.

00:19:38.543 --> 00:19:42.043
And what makes that animal species so special?

00:19:42.583 --> 00:19:48.563
Well, the New Caledonian crow is, and I'm now including the great apes,

00:19:48.623 --> 00:19:57.343
is the most intensely dependent tool user among non-humans that we know of.

00:19:58.323 --> 00:20:04.763
They show a species-wide tendency to use tools. They use it in nature.

00:20:05.423 --> 00:20:08.563
They fulfill an important part of their ecology. technology and

00:20:08.563 --> 00:20:11.603
they don't use it in a rigid fashion

00:20:11.603 --> 00:20:14.663
we know that it's not just that they have inherited

00:20:14.663 --> 00:20:17.683
a set of motor patterns they have an

00:20:17.683 --> 00:20:21.683
inherited predisposition to apply

00:20:21.683 --> 00:20:29.803
tools to problems they face and in some respects as juveniles they show a kind

00:20:29.803 --> 00:20:35.763
of proto movements of what is going to be the adult tool use and they do it

00:20:35.763 --> 00:20:39.523
for something that we might call play if we saw it in a young human.

00:20:40.223 --> 00:20:44.503
But at the same time, they have the capability for cultural transmission so

00:20:44.503 --> 00:20:50.063
that we know that they can learn from others to some extent.

00:20:50.303 --> 00:20:55.003
They have regional variation in the tools that they are using,

00:20:55.063 --> 00:20:59.723
and that could be connected to a physical culture, although that hasn't been demonstrated yet.

00:21:00.043 --> 00:21:07.443
So all these things make them very interesting. But a typical example would

00:21:07.443 --> 00:21:08.803
be that their friends would take a stick.

00:21:09.565 --> 00:21:15.985
To fish, as you described yourself, for larvae in tree, in the mask of trees, right?

00:21:16.025 --> 00:21:18.485
This would be the typical thing, how they would use a tool.

00:21:19.065 --> 00:21:24.665
Yes. Actually, local people, non-scientists, knew this for a long time.

00:21:24.885 --> 00:21:30.585
And Gavin Hunt, working from New Zealand, brought it to the attention of science

00:21:30.585 --> 00:21:39.125
in the late 90s and published some very interesting papers actually describing what they do in nature.

00:21:39.565 --> 00:21:46.825
And they use different kinds of tools, at least three or four different kinds of tools,

00:21:47.025 --> 00:21:52.305
five maybe, depending on how you categorize a kind of tool, like hooks or straight

00:21:52.305 --> 00:21:56.505
sticks or blades of pandanus plants,

00:21:56.725 --> 00:22:05.765
that in all cases, they build themselves and they modify and then they use for extracting food.

00:22:06.685 --> 00:22:11.225
We don't know to what extent in nature they are selective of producing tools

00:22:11.225 --> 00:22:16.765
that are appropriate for a particular problem that they are facing.

00:22:17.025 --> 00:22:20.725
But we know that by taking them to the laboratory and giving them different

00:22:20.725 --> 00:22:22.305
problems, they can do this.

00:22:22.425 --> 00:22:26.905
They can build tools which are appropriate for the problem they have in hand.

00:22:27.065 --> 00:22:31.725
But now if you say that they have about five tools they would use, this is like a stick?

00:22:32.045 --> 00:22:35.985
I said five categories of tools. Yes, because they're very different, all of them.

00:22:36.025 --> 00:22:44.845
But for example, they can use segments of a flexible vine that they cut with

00:22:44.845 --> 00:22:50.325
the appropriate length and poke it into holes with the teeth facing backwards

00:22:50.325 --> 00:22:52.465
so that they can actually rake things out.

00:22:52.465 --> 00:22:56.025
They can produce hooks by

00:22:56.025 --> 00:22:59.085
sculpturing twigs at the branching point

00:22:59.085 --> 00:23:02.805
and so leaving just a short segment

00:23:02.805 --> 00:23:06.205
on one side and a longer one in the other and we

00:23:06.205 --> 00:23:08.965
know in the laboratory but not in nature how

00:23:08.965 --> 00:23:12.025
these hooks are actually used they build

00:23:12.025 --> 00:23:17.605
the most complex tool of

00:23:17.605 --> 00:23:21.745
any animal really

00:23:21.745 --> 00:23:24.605
alien invertebrate if you exclude things like spider webs

00:23:24.605 --> 00:23:28.225
by actually cutting the

00:23:28.225 --> 00:23:31.485
edge of blades of leaves

00:23:31.485 --> 00:23:38.705
of pandanus trees which are like these leaves are like flat surfaces and they

00:23:38.705 --> 00:23:44.065
cut the edge with a particular step shape format so that they are thicker more

00:23:44.065 --> 00:23:49.585
robust on one side and they taper to the other and they use this for.

00:23:50.607 --> 00:23:53.947
Extraction purposes but we don't know for example why in

00:23:53.947 --> 00:23:57.227
some areas they do it and in others they don't whether it

00:23:57.227 --> 00:24:00.227
is and also why their shape is different whether

00:24:00.227 --> 00:24:06.227
it's functionally different or it's just cultural drift that leads honey that's

00:24:06.227 --> 00:24:08.967
what i was word so so interested in so if we have these different categories

00:24:08.967 --> 00:24:14.727
of tools now if you talk about cultural variation what should i think of given

00:24:14.727 --> 00:24:19.447
these categories of tools what's the culture variation here cultural variation is for For example,

00:24:19.607 --> 00:24:25.267
this is again work by our colleagues in New Zealand, that they found that the

00:24:25.267 --> 00:24:28.007
typical pandanus leaves that they produce,

00:24:28.267 --> 00:24:35.007
in some cases have a perfectly rectangular shape, which is kind of broad.

00:24:35.287 --> 00:24:41.667
In others, they have a thin one. And yet in other places, they have a stepwise

00:24:41.667 --> 00:24:47.167
one where they start thick and they end fine, which is a more advanced, more useful tool.

00:24:47.167 --> 00:24:51.427
In some areas, the three kinds of tools are built, which is,

00:24:51.507 --> 00:24:54.927
in a sense, a contradiction with the notion that there is a better,

00:24:55.067 --> 00:24:56.767
there's a best kind of doing it.

00:24:56.847 --> 00:25:04.627
But definitely what is the case is that the predominant kind of pandanus leaf

00:25:04.627 --> 00:25:08.067
tool is different in different regions.

00:25:08.227 --> 00:25:15.367
Now geographic variation does not prove culture, but it's an obvious associated concept.

00:25:16.417 --> 00:25:20.137
I was wondering, do the birds show anticipation?

00:25:20.497 --> 00:25:23.197
So if they're going to go on a particular kind of foraging trip,

00:25:23.497 --> 00:25:25.697
will they go and make a tool and take it with them?

00:25:25.957 --> 00:25:28.777
And also, do they store tools and reuse them?

00:25:29.197 --> 00:25:34.217
Well, we have shown through the use of cameras.

00:25:35.257 --> 00:25:39.477
With some of my colleagues, particularly Christian Roots, who's now in Scotland.

00:25:39.717 --> 00:25:46.997
Used to be in Oxford at the time, We managed to put cameras on birds which are

00:25:46.997 --> 00:25:50.517
moving freely in the woods and could see some of that.

00:25:50.657 --> 00:25:56.897
For example, we could see that they could make a tool and take it hundreds of

00:25:56.897 --> 00:25:59.577
meters away somewhere else and use it.

00:25:59.717 --> 00:26:05.517
But that doesn't give us any evidence that the animal knows what kind of problem

00:26:05.517 --> 00:26:09.077
it's going to face when it arrives and is doing the right kind of tool here.

00:26:09.077 --> 00:26:13.737
But in the laboratory, we know that when they face a problem,

00:26:13.837 --> 00:26:18.237
for example, where they need a hook to extract food, they actually make a hook

00:26:18.237 --> 00:26:21.017
when some other shape of tool would not work.

00:26:21.157 --> 00:26:24.317
So they really seem to be able to do this kind of planning.

00:26:24.317 --> 00:26:30.637
Similarly, when we give them a task in which they need to collect one kind of tool,

00:26:30.857 --> 00:26:35.677
say of a given length, to pick up another one that can be used to pick up yet

00:26:35.677 --> 00:26:39.837
another one, that kind of complication, then they are capable of doing it in the lab.

00:26:39.957 --> 00:26:44.937
And if you think of how you would program an autonomous machine to do this,

00:26:45.057 --> 00:26:53.257
you couldn't do it without some kind of building enrichment of the reinforcement experience.

00:26:53.257 --> 00:26:58.257
Just purely repeating what works doesn't take you to the end because the animal

00:26:58.257 --> 00:27:02.557
has never, these are trial unique things where they've never done these complete

00:27:02.557 --> 00:27:06.277
sequences. So they have to do something equivalent to what one would call planning.

00:27:06.717 --> 00:27:14.137
But I resist as much as I can to use words which are heavily laden with meaning,

00:27:14.197 --> 00:27:18.257
like planning or understanding or insight,

00:27:18.457 --> 00:27:22.437
when all they are doing is calming our anxiety about not knowing what the animal

00:27:22.437 --> 00:27:23.757
is doing by putting it a label.

00:27:24.237 --> 00:27:28.797
So, yes, they do some planning in terms of the behavior they do anticipates

00:27:28.797 --> 00:27:32.157
the problem rather than acting by consequences.

00:27:32.157 --> 00:27:35.477
Before we sort of get into the planning bit,

00:27:35.817 --> 00:27:41.597
I mean, you have gone through quite a series of experiments to really document

00:27:41.597 --> 00:27:47.717
in detail the tool use that these crows are capable of and how they can generalize.

00:27:47.717 --> 00:27:54.177
So what are the key features of their tool use that stand out in your mind?

00:27:54.697 --> 00:28:00.297
Well, the first one, and I was anticipating a second ago, is that not only...

00:28:00.933 --> 00:28:06.393
Everything they do is contained, in any obvious way, in the experience they had before.

00:28:06.773 --> 00:28:11.533
It does appear as if we need some kind of model by which the animal does what,

00:28:11.693 --> 00:28:15.373
let's say, the person in the street would call understanding of the problem,

00:28:15.473 --> 00:28:19.653
and I don't have a better label for it, but at the same time,

00:28:19.653 --> 00:28:22.233
I'm aware of the difficulties of using such term.

00:28:22.433 --> 00:28:27.533
The animal sees a problem and produces a solution that it has never experienced before.

00:28:28.113 --> 00:28:32.593
It's not just repeating with greater frequency what has worked.

00:28:32.753 --> 00:28:36.693
But on a trial-unique basis, it's solving problems one after another.

00:28:36.913 --> 00:28:41.833
That's a common feature. At the same time, we find that they need knowledge,

00:28:42.073 --> 00:28:45.793
which in many cases is completely logical to solve it.

00:28:45.993 --> 00:28:49.873
I can give you one study, for example. It

00:28:49.873 --> 00:28:57.693
was shown in other corvids by colleagues in Cambridge that rooks are capable

00:28:57.693 --> 00:29:04.133
of discovering how to drop stones in an instrument to dislodge a magnetically

00:29:04.133 --> 00:29:06.573
held platform to release a reward. world.

00:29:07.593 --> 00:29:14.933
So that was extraordinary in itself. But the Cambridge rogues had done this

00:29:14.933 --> 00:29:19.413
task by being first trained to drop stones accidentally.

00:29:19.733 --> 00:29:24.453
They could see that and then they could innovate by picking up stones from a

00:29:24.453 --> 00:29:25.953
distance and bringing it there.

00:29:26.313 --> 00:29:34.073
So what we wonder is how could these These animals know that a stone would actually

00:29:34.073 --> 00:29:38.593
dislodge the magnet without having experience of how the apparatus worked.

00:29:39.513 --> 00:29:41.813
So what we did was to...

00:29:42.830 --> 00:29:48.010
Give different groups of animals either experience with dislodging the magnet

00:29:48.010 --> 00:29:51.290
by pecking directly at it or not.

00:29:51.790 --> 00:29:57.390
And we placed them without any previous experience of stones dropping in front of the machine,

00:29:57.550 --> 00:30:03.470
and the ones that knew how the magnetic box operated could solve the problem,

00:30:03.630 --> 00:30:06.890
could innovate by bringing the keys like the rooks had done.

00:30:07.050 --> 00:30:11.170
But the ones that didn't know that contingency just looked at it and couldn't

00:30:11.170 --> 00:30:16.790
do it. So you can innovate, but you have to build on partial knowledge.

00:30:17.130 --> 00:30:23.490
Would you call it scaffolding? I would call it that, but self-scaffolding in

00:30:23.490 --> 00:30:28.610
the sense that the animal is constructing on the basis of partial elements of behavior,

00:30:28.790 --> 00:30:32.770
which is perhaps not surprising because animals don't have any reason to understand

00:30:32.770 --> 00:30:37.310
a machine without having some possibility of interacting with it,

00:30:37.390 --> 00:30:39.170
and in this case, a delivery box.

00:30:39.170 --> 00:30:43.270
But now, don't we also face a challenge here because we look at the task and

00:30:43.270 --> 00:30:44.210
it looks really complex.

00:30:44.810 --> 00:30:48.490
But in some sense, there's also an invariant in all these tasks because it's

00:30:48.490 --> 00:30:55.270
always retrieving a food item from, let's say, a tube-like structure, right?

00:30:55.350 --> 00:30:58.410
Which is in some sense a condition for which these animals might have been optimized

00:30:58.410 --> 00:31:01.590
because that's how they have to eat, find their food in the trees where they live.

00:31:02.650 --> 00:31:07.030
So maybe to us it looks very complex. but maybe for these crows it all looks

00:31:07.030 --> 00:31:11.790
like the same problem that they always solve in the same way which is get a stick, get a stick like,

00:31:12.492 --> 00:31:16.112
get a stick-like object, and start poking in that hole.

00:31:16.652 --> 00:31:21.412
No, this is not an accurate description of what's going on.

00:31:23.292 --> 00:31:28.832
I have different lines of argument here. One is that not all our experiments

00:31:28.832 --> 00:31:30.112
are about extracting food.

00:31:30.332 --> 00:31:35.592
In some of them, we give the animals unknown, potentially threatening objects,

00:31:36.492 --> 00:31:42.412
and rather than touching them immediately with their beaks, they pick up a stick

00:31:42.412 --> 00:31:43.612
and touch them at a distance.

00:31:43.772 --> 00:31:48.872
So they use tools to acquire knowledge about the world in which they are,

00:31:48.992 --> 00:31:53.352
in addition to using them to extract food. That's one line of argument.

00:31:53.812 --> 00:31:58.672
Another is that the tasks actually are extremely different in that some of them

00:31:58.672 --> 00:32:03.172
require selection of the right tool that would go through a hole, for example.

00:32:03.172 --> 00:32:10.272
We give them food in a tube which which has different holes and natural twigs

00:32:10.272 --> 00:32:16.252
that have to be sculptured to the right diameter in order to pass through the hole.

00:32:16.572 --> 00:32:22.072
They can do that. Or they have to choose the right length. Or they have to push as opposed to pull.

00:32:22.612 --> 00:32:24.712
And they learn these kind of things.

00:32:25.732 --> 00:32:31.252
There are many, many different topographies of the problem that they face.

00:32:31.252 --> 00:32:37.172
Alex, I could still argue for the sake of it that each of these examples you

00:32:37.172 --> 00:32:41.412
give me, I could decompose in terms of a stereotype behavioral pattern that

00:32:41.412 --> 00:32:43.692
is modulated in some way, right?

00:32:43.732 --> 00:32:47.972
I could say, well, also during evolution, they have learned to not approach

00:32:47.972 --> 00:32:50.332
snakes too quickly, so they always use a stick to do that.

00:32:50.452 --> 00:32:54.132
But that's a very discrete, well-defined situation where they do it.

00:32:54.352 --> 00:32:59.572
So what is the common feature of all these tasks that makes you believe that

00:32:59.572 --> 00:33:05.452
you really have to think about a fairly rich internal model and insight, right?

00:33:05.492 --> 00:33:10.352
As opposed to, let's say, also if you want a more behaviorist decomposition

00:33:10.352 --> 00:33:13.132
in more stereotyped reactive behaviors.

00:33:14.032 --> 00:33:19.572
Well, I'm hesitant about this question you're asking because on one hand,

00:33:19.632 --> 00:33:24.432
you told me that all these tasks are very similar.

00:33:24.592 --> 00:33:29.232
And then you say that they all decompose onto many different elements.

00:33:29.372 --> 00:33:34.692
I wonder what kind of human behavior cannot be decomposed also in that particular way.

00:33:34.792 --> 00:33:37.712
It's a typical behaviorist challenge, right? Yeah. Sure. Yeah.

00:33:37.772 --> 00:33:43.132
And so I think I would start by saying that the very diversity of problems that

00:33:43.132 --> 00:33:49.272
they can solve and the fact that they can solve trial-unique problems with different elements.

00:33:49.987 --> 00:33:56.047
Components, is an indication that the model we need is richer.

00:33:56.507 --> 00:34:00.887
I'm still completely in agreement with you, and I said that from the beginning

00:34:00.887 --> 00:34:07.627
today, that I'm not after finding a nucleus that we are going to call mind and

00:34:07.627 --> 00:34:09.507
we're going to say, we can't go no further.

00:34:09.867 --> 00:34:14.327
This is it. We come to this point and from then on the animal thinks.

00:34:14.687 --> 00:34:17.447
So that's not what I hope to do.

00:34:17.507 --> 00:34:23.467
What What I hope to do is to find a sufficiently rich description of the behavior

00:34:23.467 --> 00:34:28.667
of the animals and the problems they solve, that we do it algorithmically.

00:34:28.707 --> 00:34:33.547
I don't think that the basic principles of combining associative learning,

00:34:33.587 --> 00:34:38.007
all forms, you know, instrumental and Pavlovian conditioning,

00:34:38.147 --> 00:34:41.007
are going to be sufficient for this.

00:34:41.507 --> 00:34:45.707
But we know that already. We know that rats learn about the temporal location

00:34:45.707 --> 00:34:49.267
of stimuli even if they precede contingencies.

00:34:49.327 --> 00:34:55.507
We know that the pigeons can form concepts of very high level.

00:34:55.607 --> 00:35:02.007
They can decode, for example, some of the Japanese work,

00:35:02.267 --> 00:35:07.887
Watanabe and others, that show that they can distinguish a cubist from an impressionist

00:35:07.887 --> 00:35:11.467
painting, painting, even with paintings that have never seen before,

00:35:11.687 --> 00:35:15.287
simply by being trained with those categories of stimuli.

00:35:15.487 --> 00:35:19.107
Or we can think of the classic Tolman experiments around the cognitive map,

00:35:19.227 --> 00:35:22.347
showing latent learning of very complex internal representations.

00:35:22.887 --> 00:35:28.647
Exactly, absolutely, yes. As you know, some of the people who have developed.

00:35:29.534 --> 00:35:32.914
More sophistication in reinforcement learning, like Satan and Bartholomew and

00:35:32.914 --> 00:35:37.814
that kind of people, have actually shown that some of the Ptolemian type of

00:35:37.814 --> 00:35:44.434
learning could be reproduced by some enhancement to reinforcement learning.

00:35:44.854 --> 00:35:51.254
But my personal experience is, yes, some can, but some is not quite plausible

00:35:51.254 --> 00:35:55.414
in terms of being the way the animal does it. But I don't have a better alternative.

00:35:55.974 --> 00:36:01.854
Okay. I think to some extent in psychology and animal behavior,

00:36:02.214 --> 00:36:08.014
there is a little bit of an assumption that a relatively easy thing to do is

00:36:08.014 --> 00:36:11.514
association learning and a relatively hard thing to do is planning.

00:36:12.374 --> 00:36:16.554
And that kind of came out of behaviorism. And then the criticism was that,

00:36:16.634 --> 00:36:20.094
well, all these great human behaviors that we do can't be explained by that.

00:36:20.094 --> 00:36:24.994
But then in AI, people discovered very quickly that actually some aspects of

00:36:24.994 --> 00:36:26.774
planning are relatively easy to do.

00:36:27.014 --> 00:36:32.074
So things like means-end analysis was an early discovery by Herb Simon.

00:36:32.234 --> 00:36:36.054
You could program this in a computer, as long as you had the right decomposition

00:36:36.054 --> 00:36:39.254
of the problem space, which we can come back to. too.

00:36:39.874 --> 00:36:43.634
But I mean, so coming at this from the point of view of robotics,

00:36:43.974 --> 00:36:48.734
it's sometimes a little bit hard for me to understand why biologists are so

00:36:48.734 --> 00:36:54.874
reluctant to say, well, means-end analysis is something that an animal brain might be doing.

00:36:54.954 --> 00:36:58.814
Because from the point of view of the computational problem,

00:36:59.114 --> 00:37:03.154
it's perhaps not such a difficult challenge as some of the other things that

00:37:03.154 --> 00:37:04.174
our brains do really well.

00:37:04.994 --> 00:37:08.654
Yes, I agree with you. But I think that when you say biologists,

00:37:09.174 --> 00:37:15.434
I think that some biologists are excessively tolerant of such explanations,

00:37:15.474 --> 00:37:20.174
and other biologists are excessively reluctant to use them.

00:37:20.514 --> 00:37:27.134
And I'm not saying that I'm in the just middle, because I couldn't possibly

00:37:27.134 --> 00:37:29.594
say that. But the point is to add.

00:37:30.353 --> 00:37:37.373
Each process to the extent that you can test it and test its properties experimentally

00:37:37.373 --> 00:37:39.473
and model it and that kind of thing.

00:37:39.613 --> 00:37:46.613
So associative learning for me is a good candidate on the first line of battle,

00:37:46.773 --> 00:37:49.513
not because it's easier necessarily.

00:37:49.753 --> 00:37:53.373
Because in many cases we don't understand how some of this happened,

00:37:53.453 --> 00:37:58.893
but because we know a lot about it and we know that it explains a lot of the data satisfactorily.

00:37:58.913 --> 00:38:02.653
You can get the right parameters, you explain many different kinds of things,

00:38:02.773 --> 00:38:06.753
while planning or

00:38:06.753 --> 00:38:10.453
insight are descriptive processes that

00:38:10.453 --> 00:38:13.313
don't by themselves link to anything that you

00:38:13.313 --> 00:38:17.133
can actually immediately quantify and simulate

00:38:17.133 --> 00:38:21.433
as a process or write in the form of code and because

00:38:21.433 --> 00:38:25.993
of that in as much as a hypothesis has

00:38:25.993 --> 00:38:29.813
a reassuring ring to it i don't

00:38:29.813 --> 00:38:32.813
favor it as a first thing i would like i prefer

00:38:32.813 --> 00:38:38.493
hypotheses that actually stick their neck out and say well if it is this then

00:38:38.493 --> 00:38:43.073
this is what you should be seeing and so if you look at some of the early ai

00:38:43.073 --> 00:38:48.193
which was doing this sort of search space techniques and uh means sense analysis

00:38:48.193 --> 00:38:49.893
and people People like Herb Simon were saying,

00:38:49.993 --> 00:38:53.853
well, look, let's see how people, not Herb Simon,

00:38:53.973 --> 00:38:56.973
but experimentalists saying, let's look at how people solve,

00:38:57.033 --> 00:38:58.473
say, the missionary and cannibals problem,

00:38:58.693 --> 00:39:06.033
which is a means-ends analysis type task where you have to solve a difficult

00:39:06.033 --> 00:39:09.333
leap, which involves not taking the obvious next step.

00:39:09.333 --> 00:39:15.353
And so there are experimental techniques there for trying to identify when people,

00:39:15.513 --> 00:39:18.913
humans, are using these kind of planning techniques.

00:39:19.253 --> 00:39:26.153
So I'm just trying to wonder whether there is maybe more a positive agenda.

00:39:26.960 --> 00:39:32.020
That we could have for you know let's assume more of about bird cognition and

00:39:32.020 --> 00:39:38.580
see if we can design experiments that that really push uh push the possibilities

00:39:38.580 --> 00:39:40.820
of what they're doing in terms of.

00:39:42.100 --> 00:39:45.380
Identifying how there are similarities between what they're doing and maybe

00:39:45.380 --> 00:39:50.300
what humans are doing an example might be uh analogical reasoning you know can

00:39:50.300 --> 00:39:55.900
you can you develop a way of designing, maybe this has been done already,

00:39:56.140 --> 00:40:00.620
you design one apparatus which involves a sequence of moves,

00:40:01.340 --> 00:40:04.660
that the bird might discover,

00:40:05.200 --> 00:40:08.920
and then you design something which is visually very different,

00:40:09.040 --> 00:40:12.800
but there's some deep homology between the sequence of actions that you did

00:40:12.800 --> 00:40:14.260
over here that you can use over here.

00:40:14.720 --> 00:40:18.380
I mean, has that been done, or is it realistic to think about trying to do that?

00:40:18.380 --> 00:40:23.680
Well, it has been claimed to have been done in some cases.

00:40:24.660 --> 00:40:31.240
I am completely sympathetic to your view that we should have a positive agenda

00:40:31.240 --> 00:40:37.860
in that respect and trying to see, well, if the animals are using what in AI we call planning, say,

00:40:38.040 --> 00:40:42.300
then we should see these properties or we should also be capable of doing these

00:40:42.300 --> 00:40:43.940
tasks that we wouldn't have thought otherwise.

00:40:43.940 --> 00:40:52.480
So, it's not that this hypothesis or even things like, for example,

00:40:52.500 --> 00:40:54.900
episodic memory or that kind of property,

00:40:55.320 --> 00:40:59.040
I don't think that they should be left as a last resort, but they should be

00:40:59.040 --> 00:41:04.380
used when they can make new predictions that actually enrich the way you tackle

00:41:04.380 --> 00:41:06.060
experimentally the problem.

00:41:06.360 --> 00:41:10.020
So, for analogical reasoning, in some cases.

00:41:11.566 --> 00:41:15.646
There have been experiments, I wouldn't go now into the details,

00:41:15.826 --> 00:41:18.906
where certain tasks have been solved by animals.

00:41:19.166 --> 00:41:25.046
And the immediate claim has been that this actually is compatible with the animals

00:41:25.046 --> 00:41:27.866
using analogical reasoning here.

00:41:28.106 --> 00:41:32.106
Well, yes, in many cases they are compatible, but they are also compatible with

00:41:32.106 --> 00:41:34.526
other things that we understand better.

00:41:34.766 --> 00:41:37.946
It doesn't mean that they are more likely to be used by the animals,

00:41:38.106 --> 00:41:44.226
but we understand it better. So, I am interested, for example,

00:41:44.226 --> 00:41:49.526
in trying to see whether one could understand the development of these capabilities.

00:41:50.986 --> 00:41:56.886
Because, for instance, when animals resolve novel problems,

00:41:57.126 --> 00:42:02.446
they very clearly resort to some library, some database of experience that they

00:42:02.446 --> 00:42:06.566
have that has been informally acquired through interaction with the different

00:42:06.566 --> 00:42:08.346
affordances of their world.

00:42:08.346 --> 00:42:10.926
And we are testing adult animals.

00:42:11.146 --> 00:42:15.086
We don't know how they acquired that sort of database of relationships.

00:42:15.366 --> 00:42:17.906
And then they come and solve something de novo.

00:42:18.286 --> 00:42:25.126
So how do they do it? Is it because they understand what's going on at the physical

00:42:25.126 --> 00:42:31.126
level or is it simply because they are generalizing from their long-term experience?

00:42:31.126 --> 00:42:34.086
Is but not to so before we now

00:42:34.086 --> 00:42:38.406
solve this this dilemma between let's

00:42:38.406 --> 00:42:43.586
say the decomposition view or the more representation list internal view maybe

00:42:43.586 --> 00:42:46.606
we can look at a bit more the experiments that you have performed and one of

00:42:46.606 --> 00:42:51.666
them which i thought was very interesting was where you compare um the crows

00:42:51.666 --> 00:42:56.086
with with these chaos this is a parrot from from new zealand right,

00:42:57.346 --> 00:43:02.566
where you also actually got some ideas about the enormous individual differences

00:43:02.566 --> 00:43:06.086
and I think that's another aspect we should not lose sight of when we try to

00:43:06.086 --> 00:43:09.726
make some categorical decision on okay they solve it like this or like that because.

00:43:11.134 --> 00:43:16.914
If it's true that there's a large individual variability, it could even happen

00:43:16.914 --> 00:43:20.774
that within the species you cover this whole range and there's not an exclusive

00:43:20.774 --> 00:43:24.134
choice to be made, right? Why not? This is also a possibility.

00:43:24.514 --> 00:43:28.774
So in these experiments, in this comparative experiment in the Krauss and the

00:43:28.774 --> 00:43:31.434
Kess, what did you really observe? What were the key observations there?

00:43:32.554 --> 00:43:38.194
Well, what we did in that particular example that you referred to is offer a

00:43:38.194 --> 00:43:45.254
battery of different tests and trying to see whether the capacity to jump from

00:43:45.254 --> 00:43:51.654
one form of solution to the next and discovering the next one was characteristics of the species.

00:43:51.654 --> 00:43:54.314
And we did find that to some extent.

00:43:54.554 --> 00:44:00.554
But on the other hand, when we examine the reasons for success in certain tasks,

00:44:00.774 --> 00:44:08.994
we found that what you have to attribute to understanding of the physical features

00:44:08.994 --> 00:44:14.134
of the problem could be severely restricted by other differences between the individuals.

00:44:14.134 --> 00:44:20.454
For example, their proclivity to explore the environment in either a tactical

00:44:20.454 --> 00:44:24.874
modality, sorry, a haptic modality, or a visual modality.

00:44:24.994 --> 00:44:30.254
Because certain problems cannot be discovered visually, and others,

00:44:30.394 --> 00:44:35.834
which involve action at a distance, are more difficult to discover by touching alone.

00:44:36.737 --> 00:44:43.637
And differences which one would call non-cognitive get amplified into difference

00:44:43.637 --> 00:44:49.397
in problem solving and tasks which overall are designed for their cognitive interest.

00:44:49.577 --> 00:44:52.757
Because these chaos are more haptic, right, in their interaction with the world.

00:44:52.797 --> 00:44:55.297
Yes. While the crows are a bit more at the distance.

00:44:55.557 --> 00:44:59.857
That's right. At the distance. Yeah. But then, so, in my mind,

00:44:59.877 --> 00:45:03.997
surprisingly, you found these chaos were overall better in this battery of tasks

00:45:03.997 --> 00:45:07.657
than the crows. So what made exactly that distinction?

00:45:08.297 --> 00:45:16.197
Yeah, well, the crows were better at the task that they are well known for,

00:45:16.437 --> 00:45:20.977
which is to use objects to act on target at a distance.

00:45:21.037 --> 00:45:25.677
Basically, to use a tool to retrieve the, although the task was new,

00:45:25.897 --> 00:45:28.937
it had the basic structure of what they are known to be good at.

00:45:28.937 --> 00:45:38.617
But the Chias could do better any task in which direct manipulation of the object

00:45:38.617 --> 00:45:43.497
with their beaks and feet was an advantage.

00:45:44.117 --> 00:45:51.617
And if you were to take naively the results of the experiment in a quantitative

00:45:51.617 --> 00:45:56.797
way, yes, the Chias scored more highly than Euclidean crows.

00:45:56.797 --> 00:46:05.057
But we have left behind, I hope, long ago, this notion of a scalar categorization

00:46:05.057 --> 00:46:11.017
of intelligence between species because they are designed to solve different problems.

00:46:11.157 --> 00:46:19.097
And they basically have different capabilities and motivational differences cause that.

00:46:19.457 --> 00:46:23.617
I should say that you mentioned this is a very important problem when you're

00:46:23.617 --> 00:46:24.977
dealing with intelligent animals.

00:46:25.077 --> 00:46:31.157
I'm using the term here loosely to say one that uses a lot of its own experience to solve problems.

00:46:32.317 --> 00:46:37.437
And almost by very definition from the beginning of the task,

00:46:37.557 --> 00:46:40.377
you are expecting different personal histories,

00:46:40.597 --> 00:46:44.437
different individual histories to accumulate and lead to different adult behavior

00:46:44.437 --> 00:46:50.557
because nobody can guarantee what individuals are going to do.

00:46:50.557 --> 00:46:56.137
So if you look at the diversity of skills in humans and you ask who ends up

00:46:56.137 --> 00:47:01.917
being a professor of Chinese and who ends up being an engineer designing robots,

00:47:02.117 --> 00:47:06.977
for example, or a neuroscientist, it's very difficult to know what was inherently

00:47:06.977 --> 00:47:12.077
different in the alleles of different genes that these individuals had and what

00:47:12.077 --> 00:47:17.877
simply were contingencies in their personal experience that have not been formalized and cannot be.

00:47:18.457 --> 00:47:21.097
The same happens with intelligent animals. They have... Sure.

00:47:21.197 --> 00:47:24.057
But what I found interesting about your interpretation of this difference or

00:47:24.057 --> 00:47:28.317
similarity in the chaos and the cross, where you say, look, they have been specialized

00:47:28.317 --> 00:47:32.617
for different kinds of tasks, so differences should not be over-interpreted.

00:47:32.897 --> 00:47:37.697
But if I look at your result, my interpretation, I could give you an alternative

00:47:37.697 --> 00:47:41.597
interpretation, which actually, they are able to solve the same tasks.

00:47:41.917 --> 00:47:46.057
And in the Simon and Newell view, that might be the same expression of a general

00:47:46.057 --> 00:47:49.277
intelligence, if you want, where anything can become a task.

00:47:49.477 --> 00:47:54.677
However, their performance is sort of modulated by the specifics of their skeletal

00:47:54.677 --> 00:48:00.237
muscle system, their training history, the niche that they sort of have adapted for.

00:48:00.857 --> 00:48:05.857
And for the example you showed where the chaos were pretty good in dealing with

00:48:05.857 --> 00:48:09.797
a little window that was in the index mental box, while the crows could not.

00:48:09.857 --> 00:48:12.177
And the difference being that these chaos, since they're so haptic.

00:48:12.977 --> 00:48:15.757
They chew on everything and they touch everything. So in that way,

00:48:15.757 --> 00:48:19.697
they can discover this window while this crow is just perching somewhere and

00:48:19.697 --> 00:48:22.097
looking down upon the task and never figure it out.

00:48:22.377 --> 00:48:26.677
But that would mean actually that coming from these very different morphologies

00:48:26.677 --> 00:48:34.357
as a bird, being controlled by a fairly similar brain allows them to solve the same task.

00:48:34.457 --> 00:48:37.737
It's just modulated by their physical instantiation, the bodies that they have.

00:48:37.897 --> 00:48:39.257
Is that a reasonable interpretation?

00:48:39.877 --> 00:48:44.777
I think it's perfectly possible. And this is something that we would like to

00:48:44.777 --> 00:48:48.137
tease apart in a whole program of experiments, exactly.

00:48:48.177 --> 00:48:54.977
How much of what we see is just modulation of a very general ability to say,

00:48:55.137 --> 00:49:01.817
for example, represent the problem in some symbolic kind of way that allows

00:49:01.817 --> 00:49:07.517
you to explore virtually different solutions and then implement them.

00:49:07.517 --> 00:49:13.897
And to what extent they don't have this capability. They are very dedicated.

00:49:14.217 --> 00:49:22.057
We know that humans are not completely equally capable of any content-free problems.

00:49:22.237 --> 00:49:25.477
I mean, our ability to solve logical problems. But now there's an interesting

00:49:25.477 --> 00:49:26.557
consequence, right? Because….

00:49:28.068 --> 00:49:32.688
You also used the word intelligence loosely. Yeah. And intelligence is often

00:49:32.688 --> 00:49:40.888
seen as a reasoning capability based on a mysterious g-factor that recently

00:49:40.888 --> 00:49:44.668
Adrian Owen and others have shown might decompose again in other properties

00:49:44.668 --> 00:49:46.908
such as working memory and rule learning and so on.

00:49:47.448 --> 00:49:50.668
But now we have to see that if we want to insist on the notion of intelligence,

00:49:50.868 --> 00:49:52.388
we also have to include morphology.

00:49:53.148 --> 00:49:57.028
So doesn't that actually imply that we might forget about noise of intelligence

00:49:57.028 --> 00:50:02.428
because it starts to become the whole universe well I'm we try sometimes so

00:50:02.428 --> 00:50:06.408
it is very hard to leave it behind for example,

00:50:07.868 --> 00:50:10.868
we did the exercise in our

00:50:10.868 --> 00:50:17.468
laboratory to for different periods to ban certain words and see whether we

00:50:17.468 --> 00:50:22.788
could manage that and one of the words was understanding so could we We go on

00:50:22.788 --> 00:50:29.908
in our lives replacing the word understanding by what we really mean in any way.

00:50:30.188 --> 00:50:37.948
And we fail miserably. We had to bring it back and accept to use it.

00:50:37.988 --> 00:50:41.368
It's the same as doing evolutionary biology without teleology.

00:50:41.848 --> 00:50:46.868
We know that teleology is a shorthand for the action of natural selection and

00:50:46.868 --> 00:50:47.708
the different variations.

00:50:47.708 --> 00:50:55.388
But then if you don't really say what the oak tree tries to do is to make as

00:50:55.388 --> 00:50:57.348
many acorns as it possibly can.

00:50:57.508 --> 00:51:01.888
So you're not making any subjective attribution, but you are using a notion

00:51:01.888 --> 00:51:07.688
of goal, of target, which is for natural selection and not for the individual.

00:51:07.988 --> 00:51:12.248
And the same kind of thing we need to use for animal behavior all the time.

00:51:12.248 --> 00:51:17.688
Yeah, it's also the straitjacket in which behaviorism tried to put itself without much success.

00:51:18.048 --> 00:51:22.768
That's right, yes. It's just, it ends up, it brings you to study only boring

00:51:22.768 --> 00:51:27.248
problems because they are only the problems you can formalize verbally at that time.

00:51:27.328 --> 00:51:32.348
So you have to be a little bit more loose, but at the same time not have the

00:51:32.348 --> 00:51:36.108
illusion that if you say that an animal solves a problem because it understood

00:51:36.108 --> 00:51:37.568
it, you have explained anything.

00:51:37.568 --> 00:51:40.768
Of thing right but now so another animal

00:51:40.768 --> 00:51:43.848
that that you described was was this uh the kakatu this

00:51:43.848 --> 00:51:46.488
indonesian kakatu where actually started to look at

00:51:46.488 --> 00:51:50.088
a very different aspect so first it was more like what kind of problems can

00:51:50.088 --> 00:51:54.068
they solve what kind of tools can they use and actually construct which already

00:51:54.068 --> 00:51:58.568
is amazing but now with this kakatu you went again a step further to say well

00:51:58.568 --> 00:52:05.948
how many steps can they actually chain together to solve a complex problem that they

00:52:06.028 --> 00:52:09.328
actually would never encounter in nature because you built some strange contraption

00:52:09.328 --> 00:52:12.268
that even for humans might be a bit of a challenge initially.

00:52:13.148 --> 00:52:17.968
Where they had to sort of go through different steps to get a food reward.

00:52:18.168 --> 00:52:22.248
So what was the key insight and the key motivation behind that experiment?

00:52:23.720 --> 00:52:28.040
In that particular experiment, the animals had to do a series of actions,

00:52:28.300 --> 00:52:31.480
up to five actions in this case.

00:52:31.580 --> 00:52:37.180
And we'll say actions, they were very different in what actually the motor intervention

00:52:37.180 --> 00:52:41.500
had to be before reaching a target.

00:52:41.620 --> 00:52:44.160
In this case, it was a foot reward. world.

00:52:44.300 --> 00:52:50.400
But the interesting problem was that the sequence could not,

00:52:50.480 --> 00:52:55.260
none of the components of the sequence was reinforced until the whole sequence

00:52:55.260 --> 00:52:56.820
was done and in the right order.

00:52:57.100 --> 00:53:06.020
And that means that the animal could not improve progressively by reinforcement of individual actions.

00:53:06.220 --> 00:53:14.180
But what What it could do is to improve by, in a sense, perceiving the solution

00:53:14.180 --> 00:53:15.960
of anything that shortened the

00:53:15.960 --> 00:53:19.960
chain that was a physical chain of physical devices engaging one another.

00:53:20.340 --> 00:53:25.420
Anything that make it shorter was indeed progress towards achieving the goal.

00:53:26.260 --> 00:53:31.320
And that particular experiment did not require planning as such,

00:53:31.360 --> 00:53:37.560
but it required the capture of experience with a notion of goal-directedness,

00:53:37.660 --> 00:53:40.220
of sort of trying to achieve something.

00:53:40.620 --> 00:53:45.060
But the critical part of that study was that once the animals had learned to

00:53:45.060 --> 00:53:52.920
solve the problem of the sequence, we did controls in which we removed elements internal to the chain.

00:53:53.260 --> 00:53:57.120
And so now the question was, what has the animal learned?

00:53:57.240 --> 00:54:02.800
Is it going to go to the old beginning as required, or is it going to go to

00:54:02.800 --> 00:54:08.800
the first element after the removed one, so that now the problem they face is shorter.

00:54:09.160 --> 00:54:12.620
And what we found is statistical evidence that they can do the latter.

00:54:13.494 --> 00:54:18.594
And that means that somehow, that is, they do the properly functional thing

00:54:18.594 --> 00:54:24.554
of skipping parts of the chain, which are now being rendered irrelevant by the

00:54:24.554 --> 00:54:26.034
transformation of the task. Right.

00:54:26.134 --> 00:54:32.594
And what that means is that the animals are, and now I'm going to use carefully

00:54:32.594 --> 00:54:37.394
my words, the animals are sensitive to the physical interactions between the objects.

00:54:37.394 --> 00:54:40.614
Objects and if i tended to

00:54:40.614 --> 00:54:43.474
say that they understand the physical interaction between the objects but as

00:54:43.474 --> 00:54:47.194
you see i'm avoiding it so they somehow they

00:54:47.194 --> 00:54:55.234
behave as if they could eliminate parts of the task which have become irrelevant

00:54:55.234 --> 00:55:01.874
by a modification that they can see but they haven't experienced before so in

00:55:01.874 --> 00:55:04.034
your mind would be they have some sort of model of

00:55:04.114 --> 00:55:08.874
the overall problem in which they can selectively perform adaptations.

00:55:09.094 --> 00:55:13.954
I think so. But we really, I mean, just being conservative,

00:55:14.294 --> 00:55:18.654
I mean, we don't know how they do it, but we know that we can eliminate some

00:55:18.654 --> 00:55:23.234
classes of explanations which are based on very simple… What is interesting

00:55:23.234 --> 00:55:28.634
here is that from a standard Thorndikean and also reinforcement learning point of view,

00:55:29.314 --> 00:55:33.094
which in general I find a very impoverished way to think about the world and certainly biology.

00:55:34.394 --> 00:55:39.054
You would have to think about a direct reinforcement signal to first identify

00:55:39.054 --> 00:55:41.774
a step in the chain and then to link it together.

00:55:41.974 --> 00:55:46.834
So in your task, that's not possible because I really have to go through the

00:55:46.834 --> 00:55:50.574
whole chain, through all the steps from beginning to the end before I get my

00:55:50.574 --> 00:55:52.854
reward. So then you could argue, well,

00:55:53.581 --> 00:55:58.021
There might be some backward chaining. That means I have a memory representation of this.

00:55:58.281 --> 00:56:03.361
This is one way in which I could talk. I do step A. Step A is committed to a working memory.

00:56:03.601 --> 00:56:08.801
Then I do step B, and step B is inserted in a working memory until I get my

00:56:08.801 --> 00:56:12.301
reward, and then I have all these elements in my memory because they all belong

00:56:12.301 --> 00:56:13.981
together. This is how I can solve the problem.

00:56:14.821 --> 00:56:17.721
Alternatively, you could say, well, maybe Thorndike is still right with this

00:56:17.721 --> 00:56:21.841
law of effect, but the instruction signal is not external. You don't get this

00:56:21.841 --> 00:56:24.501
food reward or the worm or whatever they get, the nut.

00:56:24.921 --> 00:56:30.001
You have like an intrinsic motivational signal. Aha, it's great to solve a problem.

00:56:30.401 --> 00:56:33.821
And that's your rewarding signal that helps you to glue the steps of the chain

00:56:33.821 --> 00:56:36.921
together. So which of these two interpretations have your difference?

00:56:38.041 --> 00:56:41.981
I'm happier with the second. I would imagine that I don't think we could make

00:56:41.981 --> 00:56:45.001
the first one work in this particular task.

00:56:45.461 --> 00:56:50.021
Because the animals cannot solve the problem backwards and they have no experience.

00:56:50.021 --> 00:56:55.181
Experience the final elements of the task at all at the time that they start

00:56:55.181 --> 00:56:56.701
working on the other side of the chain.

00:56:56.921 --> 00:57:01.921
And so a backwards reinforcement procedure would never take them to the end.

00:57:02.141 --> 00:57:08.921
So, but somehow you have to give the system a method to progress,

00:57:09.041 --> 00:57:16.361
even if it's modeling it mentally, to know what is an improvement and then build on that.

00:57:16.561 --> 00:57:20.001
You may call that reinforcement or virtual reinforcement, if you want.

00:57:20.141 --> 00:57:21.961
It doesn't involve the physical.

00:57:22.901 --> 00:57:26.981
But that's interesting, right? Because hidden in that, you have a definition

00:57:26.981 --> 00:57:32.081
or a notion of, let's say, subtask completion, right?

00:57:32.181 --> 00:57:39.001
There must be something very identifiable about the event so that they can drive this intrinsic signal.

00:57:39.421 --> 00:57:45.861
So what could that be? Well, in the case of this particular task,

00:57:45.861 --> 00:57:49.161
where you have different physical devices engaging one another,

00:57:49.361 --> 00:57:55.161
if you imagine an image representation of that in your mind,

00:57:55.321 --> 00:58:04.941
you can see that you could visualize the downstream device as being movable when something,

00:58:05.201 --> 00:58:08.381
you know, a poke in the wheel has been removed.

00:58:08.561 --> 00:58:12.501
And so if that is blocking it, if I remove it, then the wheel can turn,

00:58:12.641 --> 00:58:20.781
something like that. So you need to have a notion of that sort of physical interaction.

00:58:21.301 --> 00:58:24.881
So you're saying, this is actually quite a strong assumption,

00:58:24.981 --> 00:58:28.361
which is very interesting, because you're saying, well, actually the animal

00:58:28.361 --> 00:58:33.821
has to have some sort of understanding of the whole task,

00:58:34.061 --> 00:58:36.381
of this whole machine that it has to deal with.

00:58:37.135 --> 00:58:40.955
And then within that model, it says, aha, I did step A now.

00:58:41.455 --> 00:58:45.275
Well, it's important that in this particular task, I don't want to over-interpret

00:58:45.275 --> 00:58:49.255
the data or even give them that impression at all.

00:58:49.255 --> 00:58:57.175
In this particular test, they could progress by rattling at random everything.

00:58:57.455 --> 00:59:07.195
But whenever they achieve a movement, they have to remember it perfectly the next time to it.

00:59:07.335 --> 00:59:12.855
And so, and that way, like a ratchet, they get closer and closer so that one

00:59:12.855 --> 00:59:16.755
day they come in and do one A, B, C, D, E, and then they get the fourth.

00:59:16.755 --> 00:59:19.995
So it doesn't require that they

00:59:19.995 --> 00:59:24.195
understand the physics of it to reach the solution for the first time.

00:59:24.255 --> 00:59:31.375
However, when you modify the task, you transform it by altering the order of

00:59:31.375 --> 00:59:33.075
things or removing one thing.

00:59:33.075 --> 00:59:36.195
If that was the way they learned it,

00:59:36.315 --> 00:59:42.915
and that was all that is stored in the mind of the animal, then it would not

00:59:42.915 --> 00:59:46.375
go zoom in what is the right movement

00:59:46.375 --> 00:59:50.935
now after the transformation of the task. And that's what they do.

00:59:51.095 --> 00:59:56.395
And the implication of this is that they may achieve it by something which is

00:59:56.395 --> 01:00:01.535
not very systematic and based on understanding and the logic of physical interactions.

01:00:01.535 --> 01:00:07.375
Actions but once they have done it they are sensitive to the actual physical

01:00:07.375 --> 01:00:13.135
needs of the task to create a novel solution the next time over but imagine

01:00:13.135 --> 01:00:15.715
i build a machine where i have the same.

01:00:16.435 --> 01:00:22.515
Elements i have this this this bolt i have to turn and so on but the linking

01:00:22.515 --> 01:00:27.075
to the next step which doesn't say something they have to unplug or something

01:00:27.075 --> 01:00:30.295
has to pull is not mechanically identifiable.

01:00:30.475 --> 01:00:33.575
I do that in the background through a computer, let's say.

01:00:33.895 --> 01:00:36.035
But the order stays the same.

01:00:36.475 --> 01:00:41.455
Would you believe the cockatoo would be able to solve that problem equally well

01:00:41.455 --> 01:00:45.755
if it cannot recognize the physical linking, the mechanical linking of the steps?

01:00:48.535 --> 01:00:48.555
Um...

01:00:49.936 --> 01:00:54.256
The true answer is I don't know, but I don't think is,

01:00:54.476 --> 01:01:01.036
I think it connects with some experiments that have been done in chimps, in children,

01:01:01.216 --> 01:01:09.176
and to some extent also in birds, in which you create, you rig up devices so

01:01:09.176 --> 01:01:14.456
that the movements that you cause are physically intuitive and logical or not.

01:01:14.456 --> 01:01:18.296
And you see whether the animal has greater difficulties when

01:01:18.296 --> 01:01:21.056
the action is not what you

01:01:21.056 --> 01:01:24.296
might expect from normal physical laws so we

01:01:24.296 --> 01:01:27.216
can do some of that and we indeed we we

01:01:27.216 --> 01:01:30.356
try but i can't at the moment answer your thing maybe

01:01:30.356 --> 01:01:33.116
you should build a magical machine for the bird to take

01:01:33.116 --> 01:01:36.416
apart that's right yeah to some

01:01:36.416 --> 01:01:40.616
extent your task is complicated because there

01:01:40.616 --> 01:01:43.516
are physical manipulation aspects to it

01:01:43.516 --> 01:01:47.216
which are quite challenging for for birds um but

01:01:47.216 --> 01:01:51.676
listening to you describe it it reminded me of some you know cognitive tasks

01:01:51.676 --> 01:01:55.536
that for instance development psychologists like piaget developed his serial

01:01:55.536 --> 01:02:00.096
order task which was to take a pile of sticks of different length and and lay

01:02:00.096 --> 01:02:05.876
them out in order of ascending size and then many people have looked to that in child development,

01:02:06.036 --> 01:02:13.356
or transitive inference tasks that have been investigated in children, also in chimps.

01:02:14.176 --> 01:02:18.216
Is it, I mean, have these been investigated with birds as well,

01:02:18.256 --> 01:02:22.636
these kinds of more cognitive tasks that don't give them that physical challenge so much?

01:02:22.996 --> 01:02:30.976
Yes, yes, they have. For example, the work by Alan Camille and his colleagues

01:02:30.976 --> 01:02:38.996
on what you mentioned of transitive inference is very interesting because what they show is that,

01:02:41.418 --> 01:02:48.098
some kind of physical transitive inferences required to solve some tasks.

01:02:48.318 --> 01:02:54.158
For example, if you learn that A, I'm going to try to reconstruct now,

01:02:54.278 --> 01:03:02.178
if you have that A is bigger than, I don't want to do a misrepresentation of the task.

01:03:03.618 --> 01:03:09.418
The point I'm raising is when you compare species which have a complex hierarchical society.

01:03:09.998 --> 01:03:18.298
And some which have a more egalitarian one, what you find is that they can transfer

01:03:18.298 --> 01:03:23.698
that skill to tasks which have the same logical structure, but different content.

01:03:23.938 --> 01:03:30.438
Yeah, yeah. So rather than having a modular device just capable of solving the

01:03:30.438 --> 01:03:34.658
social problem they have, they have the general, they have developed that.

01:03:34.818 --> 01:03:40.338
Now, have they developed this because they've experienced it repeatedly and

01:03:40.338 --> 01:03:45.118
they just abstract through their history this problem that the others have not?

01:03:45.278 --> 01:03:51.898
Or do they have a pre-programmed inherited logical module that allows them to

01:03:51.898 --> 01:03:54.918
do that kind of inference?

01:03:55.238 --> 01:03:58.498
That we need to do developmental studies to test.

01:03:58.718 --> 01:04:02.578
You have to see what happens if you raise animals with different level of complexity

01:04:02.578 --> 01:04:06.418
and see what kind of logical tasks they can solve later.

01:04:06.738 --> 01:04:17.398
So now you did say that these crows use tools and cockatoos do not in the wild.

01:04:17.578 --> 01:04:18.938
Can you really be sure about that?

01:04:19.258 --> 01:04:26.078
No. Okay. I said that crows are very well known for their extensive use of tools

01:04:26.078 --> 01:04:29.358
in the wild and cockatoos are not known to use tools.

01:04:29.358 --> 01:04:36.518
Now, if they were as intense tool users as the crows are, we would know.

01:04:36.858 --> 01:04:41.198
But they have not been studied in the wild, actually, in greater detail.

01:04:41.338 --> 01:04:48.558
So there is, I would never claim lack of an ability or someone is going to find it.

01:04:48.558 --> 01:04:51.358
If they have the skill, why wouldn't they do it?

01:04:51.358 --> 01:04:56.918
In the case of capuchin monkeys, they were shown for many years to have great

01:04:56.918 --> 01:05:00.138
capability for the use of tools in the laboratory,

01:05:00.378 --> 01:05:05.338
and people kept repeating that they don't do it in the wild until some groups

01:05:05.338 --> 01:05:09.178
of researchers found them doing it in the wild and doing it a very sophisticated

01:05:09.178 --> 01:05:12.638
kind of tool, which is now a classic study.

01:05:12.638 --> 01:05:15.498
So in your torture conclusions at some

01:05:15.498 --> 01:05:18.898
point you made the point we play chess because we

01:05:18.898 --> 01:05:24.318
are bad at it so what are you trying to this is an intriguing statement the

01:05:24.318 --> 01:05:29.958
point I was trying to make is not that we play chess I was saying we are impressed

01:05:29.958 --> 01:05:37.598
by performance in chess and people use chess as I think this idea is borrowed from Chomsky,

01:05:37.638 --> 01:05:39.678
it's not my own but I just.

01:05:40.421 --> 01:05:44.221
I can't locate at the moment where I read it, but I remember it labeled clearly

01:05:44.221 --> 01:05:45.701
as coming from reading that.

01:05:46.141 --> 01:05:52.421
Basically, if you compare something like the linguistic skills of humans and

01:05:52.421 --> 01:05:57.961
the speed with which we acquired it at an early age with extreme poverty of data,

01:05:58.221 --> 01:06:02.761
although not absence of data, as it has sometimes been claimed, but poverty of data.

01:06:02.761 --> 01:06:10.781
We very quickly learn extraordinary complex tasks, while chess has a very limited set of rules.

01:06:11.001 --> 01:06:16.921
And we nevertheless don't become half as good at doing it.

01:06:16.921 --> 01:06:21.761
And there are great individual differences in chess playing ability while there

01:06:21.761 --> 01:06:28.041
are perhaps less differences or we don't use them as a measure of capability

01:06:28.041 --> 01:06:32.941
for producing normal human speech.

01:06:32.941 --> 01:06:39.001
So, all I was saying is that we have to be careful that we may end up using

01:06:39.001 --> 01:06:49.541
things which are particularly outside the range of competencies of a species as interesting,

01:06:49.741 --> 01:06:52.801
particularly because the species is bad at it.

01:06:52.941 --> 01:06:55.101
Right. Although not everyone is a poet.

01:06:55.481 --> 01:07:02.481
Not everyone is a poet, indeed. So then you made another important point in

01:07:02.481 --> 01:07:05.541
your conclusions, which had a lot to do with sort of the comparative aspect

01:07:05.541 --> 01:07:08.401
of this, like what the things we learn about birds,

01:07:08.501 --> 01:07:10.901
how would it generalize to other species?

01:07:11.101 --> 01:07:15.621
And you made this point about the relationship between body size and brain weight

01:07:15.621 --> 01:07:17.921
that you found sort of speaking to this point.

01:07:18.041 --> 01:07:23.801
So what's the message there? Yes, the point I was trying to make is we know

01:07:23.801 --> 01:07:28.801
that absolute brain size is not a very rich indicator of capability.

01:07:28.941 --> 01:07:36.321
But on the other hand, it's hard to believe that the absolute size is entirely irrelevant of a brain.

01:07:36.321 --> 01:07:45.261
But basically, by people playing around with different ways of plotting brains,

01:07:45.521 --> 01:07:51.221
different sizes of brains, scale them in different forms, they found a way in

01:07:51.221 --> 01:07:52.721
which humans do particularly well.

01:07:53.121 --> 01:07:59.701
And that has stuck. And this is actually to look at the relation between human

01:07:59.701 --> 01:08:04.421
brain size relative to what you would expect from a mammal of that size.

01:08:04.561 --> 01:08:09.121
And you find that the residual on that overall correlation favors humans very dramatically.

01:08:09.981 --> 01:08:14.141
We don't have the biggest brains in nature, but we have the biggest residual

01:08:14.141 --> 01:08:15.781
with respect to animals of our size.

01:08:16.881 --> 01:08:20.381
All I was pointing out is that if you look at the whole of birds,

01:08:20.621 --> 01:08:23.661
and particularly if you looked at passerines, The,

01:08:24.749 --> 01:08:31.489
area in this plot of brain size versus body mass that you have is very similar.

01:08:31.669 --> 01:08:35.349
So birds tend to be smaller than at least the larger mammals,

01:08:35.629 --> 01:08:39.069
but for a given size, they have similarly sized brains.

01:08:39.449 --> 01:08:43.869
But within that, what you find is that the parrots and the crows,

01:08:44.129 --> 01:08:48.829
which happen to be the groups of animals for which we have greater evidence

01:08:48.829 --> 01:08:56.649
of these cognitively difficult tasks are also the ones which have greatest residual

01:08:56.649 --> 01:08:58.609
respect to the overall regression in the birds.

01:08:58.849 --> 01:09:03.969
So all I was saying is that, yes, there are differences between species.

01:09:04.189 --> 01:09:05.649
Not everybody is equally smart.

01:09:06.049 --> 01:09:15.329
Maybe parrots and crows are really capable of greater general logical processing

01:09:15.329 --> 01:09:19.289
of the the world and they have something more of a G-factor than,

01:09:19.769 --> 01:09:25.029
others, but as a biologist, I wouldn't jump to that until I exclude the direct

01:09:25.029 --> 01:09:25.989
link with their ecology.

01:09:27.169 --> 01:09:31.609
Right. But now, with birds, also crows, they're one of the few animal species

01:09:31.609 --> 01:09:36.329
where people have shown that they do have something like self-consciousness, right?

01:09:36.369 --> 01:09:39.229
So you can stick, it's a typical task, right?

01:09:39.269 --> 01:09:42.289
You stick something to the body, you put the animal in front of a mirror,

01:09:42.389 --> 01:09:45.009
and apparently they will take it off. So.

01:09:46.066 --> 01:09:49.146
Do you buy these kind of experiments?

01:09:49.286 --> 01:09:53.366
Do you think that something like self-consciousness is important in the kind

01:09:53.366 --> 01:09:57.006
of mental operations we're talking about for these tasks, or you see this as

01:09:57.006 --> 01:09:58.006
completely irrelevant? relevant?

01:09:58.406 --> 01:10:02.846
I buy the experiments in terms of empirical results.

01:10:03.206 --> 01:10:09.746
What I don't think is fair is to jump from the ability to address behavior to

01:10:09.746 --> 01:10:13.406
your own body on the basis of external stimuli to some philosophical notion

01:10:13.406 --> 01:10:16.126
of selfness and identity.

01:10:16.566 --> 01:10:23.126
You could be trained by watching mirrors to identify self by the contingency

01:10:23.126 --> 01:10:26.386
between your movements and the stimulus that you're seeing outside.

01:10:26.746 --> 01:10:30.146
That's a little bit more difficult than looking at your own hand.

01:10:30.326 --> 01:10:36.466
You are looking at your hand in the mirror, but still you are learning from your experience.

01:10:37.146 --> 01:10:41.066
But that seems an interesting transition because with the problem-solving behavior,

01:10:41.546 --> 01:10:45.846
you were very much at the end of, okay, there's a complex internal model in

01:10:45.846 --> 01:10:46.966
which you perform operations.

01:10:47.606 --> 01:10:52.446
Well, if we now talk about self, you seem to try to reduce it away to simple

01:10:52.586 --> 01:10:54.066
externalized stereotype behaviors?

01:10:54.266 --> 01:10:58.206
Why is there not a model of the self then as well, as much as there is one of a complex task?

01:10:58.926 --> 01:11:05.546
Well, Paul, if you are insinuating that I'm contradictory, I'm quite willing to accept it.

01:11:07.686 --> 01:11:15.906
I can't help but reflecting the different realities that we observe,

01:11:16.026 --> 01:11:20.266
having a different degree of confidence in what we can infer.

01:11:21.306 --> 01:11:27.206
I don't think I think it's fair to describe my previous statements and saying that I was...

01:11:28.304 --> 01:11:34.224
Kind of defending a highfalutin interpretations whenever a simple one would do or the opposite.

01:11:34.304 --> 01:11:38.004
No, no, I wasn't trying to say that. Closer to the culture one,

01:11:38.104 --> 01:11:41.704
but often we fail and then we have to elaborate on that.

01:11:41.744 --> 01:11:46.064
My key point was in the problem solving, it's clear that you do assume that

01:11:46.064 --> 01:11:48.104
there's an internal model, which is completely reasonable.

01:11:48.564 --> 01:11:53.284
It's an internal model of the task. While in the self-oriented behaviors,

01:11:53.644 --> 01:11:56.044
you might equally say there is an internal model of self.

01:11:56.184 --> 01:12:00.104
Why not? But you seem to sort of not want to go that way.

01:12:00.244 --> 01:12:08.024
I'm hesitant because I don't find the evidence of the mirror self-recognition

01:12:08.024 --> 01:12:14.504
task sufficiently compelling to tell me that the animal has to have what we

01:12:14.504 --> 01:12:17.884
normally understand by the notion of the self, the notion of agency.

01:12:17.884 --> 01:12:28.284
I believe that it's natural that the club of those that can do it can only grow. It can't get smaller.

01:12:28.564 --> 01:12:33.864
So people start by showing that it only happens in humans, and they say that this defines humans.

01:12:34.124 --> 01:12:38.424
Then they find that this also happens in chimps, and then next they find that

01:12:38.424 --> 01:12:41.824
it also happens in some corvids, in magpies.

01:12:42.864 --> 01:12:47.144
And then they keep up, and then eventually they show that it works in fish.

01:12:47.144 --> 01:12:48.924
And now they have a problem.

01:12:49.064 --> 01:12:59.324
Do fish also have this virtue that we call identity, self-conscious, self-recognition?

01:12:59.444 --> 01:13:04.584
Or the experiment didn't require that in the first place when we saw it in humans.

01:13:05.784 --> 01:13:11.584
This is a very common process. Sure. So, Alex, to finish up, two questions.

01:13:12.604 --> 01:13:17.084
It's clear that you have a broad experience in this domain, studying animal

01:13:17.084 --> 01:13:23.264
behavior over many years, having deep insights in the capabilities of many of these animal species.

01:13:24.904 --> 01:13:31.404
So, what would be Alex's law that we should adhere to in the study of animal cognition?

01:13:33.304 --> 01:13:33.904
Hmm.

01:13:35.588 --> 01:13:41.468
I would say that to, I'm very Tim Bergen in this,

01:13:41.628 --> 01:13:51.108
and, you know, I think that the advice of, or the way that Nico Timbergen in the 60s, in 63,

01:13:51.368 --> 01:13:55.808
structured the program of ethology is still a very good program, you know.

01:13:55.808 --> 01:14:01.028
He said, when you look at any behavior, but you could apply to any biological

01:14:01.028 --> 01:14:05.688
trait, don't think that a single level of explanation is sufficient.

01:14:06.068 --> 01:14:09.668
So don't use exclusively mechanistic interpretations.

01:14:09.948 --> 01:14:15.628
This is the way they do it and satisfy with that, nor purely normative ones.

01:14:15.808 --> 01:14:20.268
This is why they do it this way and satisfy with that. I think an interaction

01:14:20.268 --> 01:14:26.948
between normative and mechanistic approaches is absolutely essential.

01:14:27.948 --> 01:14:32.108
But another thing that is essential is to use the right level of reductionism.

01:14:32.108 --> 01:14:39.028
If we were to jump now immediately with our current level of knowledge to some

01:14:39.028 --> 01:14:42.828
of the very basic properties of birds'

01:14:43.108 --> 01:14:48.008
nervous systems, I don't think we would make much progress in the kind of problems

01:14:48.008 --> 01:14:49.128
we are facing at the moment.

01:14:49.568 --> 01:14:56.348
We can still do a lot by working at the behavior of the animals and making inferences

01:14:56.348 --> 01:14:59.888
from that, taking the ecology into account, taking evolution into account.

01:14:59.888 --> 01:15:05.168
But of course, in the end, you do want to reduce one step at a time and go as

01:15:05.168 --> 01:15:09.708
far as possible to the basic machinery of how the animals achieve it.

01:15:09.848 --> 01:15:13.348
So we should follow some plurality, though, in our view on these phenomena.

01:15:13.848 --> 01:15:17.288
Absolutely. And there's one thing I said right at the end in the conversation

01:15:17.288 --> 01:15:21.948
in the question time was that I believe much more in multidisciplinarity than

01:15:21.948 --> 01:15:23.528
I believe in interdisciplinarity.

01:15:23.528 --> 01:15:28.108
What I mean by this is we can't be equally good at everything we do,

01:15:28.188 --> 01:15:34.988
but we can meet with colleagues and form teams where everybody is very good at everything.

01:15:35.615 --> 01:15:39.615
A particular way of looking at nature. And in that, you also see a clear role

01:15:39.615 --> 01:15:42.855
for computational and robotics-oriented approaches.

01:15:43.495 --> 01:15:48.455
Very much so. I really think that we are,

01:15:48.675 --> 01:15:52.575
well, not fully because here we are doing it,

01:15:52.675 --> 01:15:58.615
but we would be losing and missing opportunities if we didn't use the wisdom

01:15:58.615 --> 01:16:04.655
acquired by people in AI and robotics on what you need to have in a machine

01:16:04.655 --> 01:16:09.595
to be able to take autonomous decisions and to construct novelty in behavior.

01:16:09.795 --> 01:16:13.935
And we will not use that wisdom to interpret the behavior of animals that we

01:16:13.935 --> 01:16:18.355
see doing that, but we have no idea of what process is underlying it.

01:16:18.435 --> 01:16:23.755
So I think working together would tell robots, sorry, roboticists,

01:16:24.375 --> 01:16:30.535
what kind of problems they may aspire at solving by looking at what animals can do.

01:16:30.535 --> 01:16:35.855
And it would tell us how some of these problems can actually be tackled because

01:16:35.855 --> 01:16:39.035
they have been tackled in machines being built.

01:16:39.195 --> 01:16:43.055
So now five years from now, Tony and I will come visit you in Oxford and we're

01:16:43.055 --> 01:16:45.755
going to confront you with the prediction you're going to make today.

01:16:46.235 --> 01:16:51.595
So I'm going to ask you whether it was validated or invalidated in this intervening period.

01:16:51.895 --> 01:16:56.195
So what's the one prediction you really would like to make today and you really

01:16:56.195 --> 01:17:01.275
would like to stick to and rigorously investigated over the coming five years.

01:17:01.695 --> 01:17:04.155
Gosh, Paul, you didn't warn me of this question.

01:17:05.655 --> 01:17:12.255
Surprise, surprise. Yeah, yeah, yeah. Give me a few seconds to say something

01:17:12.255 --> 01:17:15.975
non-completely trivial. Um...

01:17:18.592 --> 01:17:22.912
I would say that certainly in my own field,

01:17:23.052 --> 01:17:28.252
if I can be a little parochial, the behavioral ecology field that evolved out

01:17:28.252 --> 01:17:34.752
of ethology by moving the swing of interpretation towards purely functional

01:17:34.752 --> 01:17:39.852
analysis would have moved considerably back in the direction of paying attention

01:17:39.852 --> 01:17:41.272
to how animals do things.

01:17:42.372 --> 01:17:48.692
But this is a prediction about science, not about animals, so I may be cheating a little bit.

01:17:49.092 --> 01:17:53.512
But I'm saying that what is going to happen is that people are going to realize,

01:17:54.552 --> 01:18:00.292
that making purely normative functional interpretations without looking at how

01:18:00.292 --> 01:18:05.752
animals actually achieve it becomes sterile in the end. So we need that.

01:18:06.552 --> 01:18:09.372
Wonderful. Alex Czelnik, thank you very much for this conversation.

01:18:09.872 --> 01:18:11.212
Thank you. It was a pleasure.

01:18:11.120 --> 01:18:18.000
Music.

01:18:39.372 --> 01:18:42.172
Go to csnnetwork.com.

01:18:43.120 --> 01:18:51.760
Music.