Yngve Holen

The camera team from the Frankfurter Allgemeine Zeitung is coming in an hour, so let’s see how much time we have.

Ok, an hour is probably good.

Let’s see how far we get. What do you want to discuss today?

Well, we want to see the monkeys. Is it possible to see the monkeys?

There are no experiments done today because the Frankfurter Allgemeine Zeitung is coming with a camera team. The animal house is closed.

Closed.

Closed. Nothing to be seen.

Okay.

I can tell you it’s boring.

Boring. Okay.

All you will see is a black painted booth where the monkeys normally sit, and they sit in the plastic chair, and they watch a monitor in front of their head. They have electrodes implanted in the brain, and then there’s a plug on the surface of the skull. They get plugged with a flexible wire, and then they sit in this chair. They have to sit in the chair because we don’t want them to have their hands free –

To take the plugs off?

That’s the only reason.

They’re like us.

So they sit there. They have buttons to press, which they can manipulate. And they look at the video monitor. We ask them to fixate on a small dot on the screen so the eyes are at rest – so that we have control over the eye movement during the whole trial. The dot appears, they fixate on the dot, and they have to remain stationary. Then we present patterns at places we preselect, places where we know the response areas of the neurons.

And what do they do?

They have several tasks, either a simple detection task, if they see a movement or a change in the pattern – just to keep their attention. Or if you want to examine detection effects, we show two patterns, so the color changes a little bit. So, for example, it’s now pink, ignore the stimulus on the left side, you have to pay atten- tion to the one on the right. Respond to a change here, respond to a change there, and it further complicates.

Pretty boring. I wonder if humans could do it. They might need ADHD drugs.

Monkeys don’t use the internet so they don’t have ADHD.

That’s true. Another task is to study memory. We show them places on the monitor, a sample on the monitor. It could be artificial, a graphic, natural images. Then you switch them off, there’s a delay. The monkey has to remember what they’ve seen. And then you show them a test picture, and they have to decide: Have I seen this? Is that from the samples or is it new? And it’s also how you arrange it. The same: press button right. Different: press button left. Or they don’t press at all and they have no chance to be rewarded.

How do you reward them?

For correct performances, they are rewarded with a few drops of fruit juice – stuff that they like. And then they work while we record activity from their brains, until they had enough.

How long do they work?

Sometimes they work three hours, four hours, and then they stop working, and sometimes they fall asleep. We then wake them up again. But if they don’t want to, they don’t have to – we don’t force them. So they go back into the animal colony and we revisit them two days later. That’s the procedure.

Animal colony?

That’s what we call where they live.

Where is that?

Here.

Can we see that?

Not today.

So they work for juice?

They work for fruit juice. After a while the well-trained monkeys get pleasure from just getting it right. We associate a tone with correct responses, and another one with incorrect responses. So they know beforehand if they got it right or not. So they know if they’re doing well or not. If they quit fixation, then the trial is aborted. Even if you don’t give rewards, they have feedback.

You’re working with the visual system right?

I take the visual cortex as a model structure, but I could as well work in the auditory cortex, or in other parts of the cortex. Assuming that the functions realized by this very special circuitry are generalizable. The visual cortex relies probably on the same computational algorithms with the information it gets as the auditory or tactile cortexes.

Why the visual cortex?

Because it’s well explored. Because we have experience with it. I worked a lot on development in the visual cortex.

Doing what?

What we try to solve is how this immense amount of information that we have stored in the brain on natural environments – partly genetically printed already because of evolution, partly acquired during early life, and partly also acquired throughout life with normal experience – how this extremely large body of knowledge is stored in the circuitry of the cerebral cortex, and how it is possible to access it so quickly. You make an eye movement every 200 milliseconds, meaning that every 200 milliseconds the sensory evidence that you get changes. And you have to match this on the fly with these stored priors, and you have to pull out the right priors in order to cement the image and identify the object.

How can this be done? What is the sto- rage space like for such a thing?

Clearly it’s not like in computers where you have it in a list or serially. Memory must be highly parallelized, you must superimpose all this information some-how, and then have rapid access to it. The hypothesis that I propose is that this can only be done if you do all these operations in a very, very high-dimen- sional-state space. For this you need high-dimensional dynamics, and there is a very pertinent structure in the cere- bral cortex. You have nodes or columns in the network made up of cells. And these cells have certain response pro- perties, they are attuned to certain fea- tures – orientation, direction, motion, color, contrast, and so forth. In some of the areas the response properties of these nodes are much more complex, represent combinations of elementary features. And these nodes, columns, or classes of cells, they’re all reciprocally coupled in the visual cortex, as well is in all the other cortical areas – it’s the same principle. These couplings decay in an exponential fashion with distance, so not everyone talks to everyone directly. To talk to someone further away you have to do it indirectly. And the very important feature of these connections is that they are adaptive, they can learn. They learn according to the well-known grouping-sensation rules, or feature- detecting neurons that have a high pro- bability of co-occurring in natural envi- ronments. They have a cell that looks in the orientation here and another one, orientation here, because there is so much collinearity in the outer world – there is a lot of order in the visual world, in the world in general. What happens is that neurons code for features that tend to co-occur very often, like oriented lines that are colinear. Or same texture here, same texture there. Or coherent motion, which is also a globally coherent pattern that sequentially activates neurons that prefer the same direction of motion.

Are these regularities in the environ- ment captured by the gestalt principles?

The gestalt psychologists have put up a whole set of principles that allow you to sort the essential evidence according to criteria of likelihood of co-occurring together, that can be bound together to form a figure to segregate from the background. So the idea we have is that... Well, there’s proof that these connections learn these contingencies, they strengthen when they exist between feature detectors that are very often co-activated in a correlated way.

What does that mean?

Meaning that those features tend to co-occur very often. So these neurons look at all the features in the scene and encode for features that are worth being bound. With all likelihood they belong to the same object because in the past they have occurred together. They then form – these neurons that are prefe- rentially coupled – they form an ad hoc ensemble of coherently active neurons that become synchronized. Much more easily than neurons that are only weakly coupled. So what you get is you have this very, very dense network of recurring connections, these reciprocal couplings between all these feature-detec- ting neurons that have learned in the past about the statistic regularities of the environment. This knowledge is now sitting in the functional architecture of these connections. It’s latently there, it’s not read out yet. The asymmetry in these couplings are the latent storage of all this knowledge that you need of these priors.

What you mean is that one side is order and one side is chaos? The input is chaos?

I should first say that during spontaneous activity you have this complex – not unstructured, but very complex – high-dimensional pattern of activity that evolves or emerges from this network. It’s as if all these priors, all this knowledge, was latently encoded to be called upon but it is not realized yet. It’s hovering around everything and superimposes very quickly. And then you get sensory evidence from the visual or tactile system or whatever. Then signals come in that match some of the in-built priors. That will drive the neurons that are preferentially coupled, and these neurons will immediately exchange their activity and become coherently active and they synchronize. And we see this is manifested in the brief oscillation in the particular frequency ranges 40 Hz, 30 Hz, gamma frequencies.

What does that do?

All of a sudden it reduces the dimensionality of this state space. There are substates that become more synchronized, less complex, more orderly, and these substates, they now represent the result of a match of the incoming sensory evidence with the already stored knowledge. And because they produce these low-dimensional synchronized soft states, they are propagated forward, and can be very easily classified. They are more consistent than what you had before.

What if you have a stimulus that has never been seen before? Something unique?

That wouldn’t have much internal structure. It will also create a substate, but it’s a substate that is much less ordered. It would cause the collapse of low-dimensionality, and it’s much more difficult to classify. This is the hypothesis we pursue. It has a little bit to do with reservoir computing or liquid computing. Echo-state computing –

Is that like neuromorphic computing?

In a sense, it is of course neuromorphic, because you’re adding neurons to it. It’s quite different from what you now read a lot about these deep learning networks that do packet classification. All these recurrent connections, they are simply feet-forward connections, many layers. They are good in classifying feature constellations, but they do not extract semantically meaningful objects, let alone relations between objects. They just classify a bunch of features. So it’s a very different principle. You find recall networks all over in the brain, in all cortical structures. You also find them in the hippocampus. You don’t find them in other structures. They are an acute invention of evolution.

Why did it evolve?

Because recurrence allows you to create these very high-dimensional dynamic states. You can imagine if element A talks to B and B talks to C and C talks back to B and to A. If you have millions of those you get a very complex pattern that produces the high-dimensionality of these states. You can’t intuitively imagine them. Some people say the dimensionality of this system is infinite. You can’t really imagine what it is. Nor can you get a good intuitive grasp of the dynamics. We talk about time being the fourth dimension. Here we’re talking about very, very, very many dimensions. It’s quite curious that you have a machine in your head that does all the stuff you know it does and you have no good intuition for the mechanisms that are underlying it.

What role do concepts play in this?

Basic. If you had no concept in mind, no working hypothesis, you would just collect data and you wouldn’t know what to do with it. The space that you can explore is really infinite, recording the activity of all those neurons. If you wouldn’t have a hypothesis, or at least an intuition or what is likely the case, you wouldn’t know what to look for. So usually this type of research is hypothe- sis-driven.

But what is a concept?

A concept?

An idea. How does it emerge?

It’s part of our ability to reason. I guess what you have to do is encode content at a certain level of abstraction so that you can establish semantic relations among the different elements of this content, following logic and principles in general. And trying to arrive at a coherent picture or interpretation of facts that you are aware of. That’s usually what we address as a concept. It should be free of contradictions and it should have explanatory value.

But how do we form this in the brain?

We have no idea. It’s probably closely related, in humans at least, since we are speaking animals, to the organization of language in our brains. But you don’t have to have these logical rules to develop a concept that allows you to say: this painting is finished. But you must have internal criteria to make that judgement, and it’s also based on a concept. Where that concept comes from is unknown. We don’t know much about that.

I first read your work in the context of the debate about free will, more than 10 years ago. It was a big topic. We were reading an old manifesto today from 2004, discussing how we know a lot about brain regions, and that we can also study small things, neurons, but what’s happening in between is completely unknown. Is this still the case?

I think the bottom line is that we have accumulated an enormous amount of new data using new technologies, but conceptually we haven’t advanced that much. We are at the turn of the moment of what we consider 20th-century neuroscience and 21st-century neuroscience – the difference being that 20th-century neuroscience was still more in the framework of cybernetics.

What do you mean cybernetics?

It was more in the framework of serial operations in the hierarchical system, that is input-driven, does something, then there’s an output. While now we see the brain much more as a very complex, self-organizing system of nonlinear dynamics, that is generative, that produces hypotheses, questions, all the time.

It talks mainly to itself?

Only a fraction of the synaptic activity in the cerebral cortex are made by input from the periphery. All the rest – 90 percent – comes from within. It is a constructive system that takes signals from the environment to confirm hypotheses rather than waiting until something happens outside and then making sense out of it. We are very convinced that perception – the way we perceive the world – is a construction that follows results from prior knowledge, from our expectancies, and from a lot of implicit, covert knowledge that we have no control of. The brain computes stuff on the basis of sensory evidence and presents this as an experience. But very often we don’t even know how it came to that conclusion.

Making the question of free will –

I think neuroscience supports constructivist philosophical stances. The free will question, in my eyes, is a trivial one. If you believe that... Unless you take a dualistic stance, and you really think about the world of consciousness and psyche and spirituality being an ontolo- gical entity apart. And then you have the material world on the other side, and the two in some mysterious way interact. Unless you defend this position, you have to assume the naturalist position that all the mental functions, including our consciousness, our feelings, et cetera are the result of neuronal interactions. If this is true, then what you do, what you decide, what you feel, what you see, must follow the laws that govern the activity in the brain. And these are the laws of nature. So causality is on important prin- ciple and obviously acts there as well. How you decide can only depend on the way your brain works, plus a little bit of serendipity, some noise. A dice sometimes falls to the right, and sometimes falls to the left.

But that doesn’t set you free.

It just makes you dependent on hazards rather than laws. It doesn’t help much. I think it’s trivial. This whole debate only got heated because people came to the wrong conclusion, which they read out of the papers. I never said this. If you are not free, in the sense that you could have done anything else, but you just did this and this was the reason, this was your free will decision, I would say, the reason for this was because the brain has a history, and it behaves according to this history, even though you may not be aware of everything that may determine such an outcome.

So if you are not free in the sense that, you could have done anything, but just did this, then you cannot be responsible for what you do.

This is of course nonsense. You are the author. Who else? It’s you. So you are responsible, which implies that society must have the right to tell you: look, what you did here is not admissible. We don’t allow you to do that. It’s what we do with our children, even though here we think they are not free, because we don’t think they are mature enough to have free will. We punish them or reward them, so I think this whole debate was the media. The hype comes and goes.

But it triggers a lot of discussion on the level of law. Some people believe that neuroscience shows that the criminal justice system makes no sense.

But we don’t know enough about the brain to abolish it. If you find a cause for an inappropriate behavior that’s neurobiologically linked, using X-rays, MRIs, whatever, like a jury, you would send the person who produced that inappropriate behavior to the clinic. If you can’t find a cause, because your measuring instruments aren’t appropriate, then that person goes to jail. This is an interesting thing. Neurobiologists would say you always have a neuronal cause for a mis- behavior. It may not be a tumor but the brain could be misfiring – there are many reasons you behave in certain ways for which you cannot see a cause from out- side. So detectability of abnormalities becomes the criteria to decide whether to send someone to the clinic or to jail. And this is a point that needs to be dis- cussed, that has been discussed, that is discussed.

Do you believe that there will be a fundamental shift in how we treat abnormal behaviors with the more data and knowledge we have about how the brain works?

Whether we can discover the causes let alone treat the abnormalities is the ques- tion. We know the causes of Alzheimer’s but we don’t know how to treat it. But yeah, maybe in the long run. I think education is an important treatment. You can change the architecture of the brain through education, that is, through experience. The brain develops until you’re 25.

And then?

Well, the process until you’re 25 is still developmental. You have new connections formed and existing connections retracted, depending on the use. You wire together what fires together. You use correlations. And this brings you to the adult architecture.
And then you have what you have, and you have to live with it. All you can do is you can still modify the connections, the efficiency of the connections. 

So you can’t make a new connection and you can’t destroy one that is already there, a disturbing one, for example? 

You can only increase or decrease the efficiency by changing the synaptic gain, which is learning. You can learn to control strange behavior that you have because of genetic wiring. You can learn to suppress it. And this plateau phase lasts until about age, probably my age, 70 or 75 or so. And then even under very normal conditions, you notice a loss of connections, a loss of synaptic connections, ultimately also a moderate loss of cells, cell number. And then, yeah, you become cognitively impaired. You become slower, less sharp, maybe wiser, because you don’t care too much about details anymore. 

What are the possibilities of shocking the brain into reconfiguring itself? Or even slow massive shifts in brain func- tion, like those you’ve discussed with the Buddhist monk Mathieu Ricard? 

We don’t know very well. If you practice a lot of meditative routines you get to know yourself better, and that allows you to perceive the world at a greater distance. You have a more objective view of the world, what perception is. If you’ve seen your internal mirror, if you wipe clean your internal stream, because you get to know yourself well, then your picture of the world becomes more realistic. And that alleviates suffering, and you become a better person. This is what Mathieu would say. To which extent this works or not –

What about too much meditation? Can that make you crazy?

I have a daughter who does real research on this. The outcome is...There is evidence... It does produce a change. To which extent this is lasting beyond the practice, I wouldn’t know. I myself did one of these crash courses in Zen meditation for a fortnight, eight hours in front of a white wall. Counting from one to 10. It did change something. I got to know a part of myself that I didn’t know before, that I can reactivate now when I sit and be quiet. It certainly did something. We also know from trauma research and from cathartic events in life that they can shock you to an extent that you are no longer the same afterwards. What that entails in terms of mechanisms I don’t know. We start to know a little bit about the consequences of prolonged stress on brain functions. And of course, a changed resilience to stress changes your behavior. But to which extent you can change the character of a personality is not known. With adult brains, through meditation, it is said that you can clean your consciousness to an extent that it becomes a reliable reflector of reality. Ultimately this would entail that there can be a conscious state without content – that you just clean, clean, clean, and then you have it, and it can just come in – I don’t know if that’s possible.

Can’t realizing emptiness or whatever lead to psychosis? Like all forms of isolation?

It’s a research question, but so far we have no empirical approach to answer it. 

How much Western scientific examination is done on these states of mind? Can you scan a monk’s brain with an fMRI?

This has been done. Rich Davidson in Wisconsin has done quite a lot on that. Other groups have taken well-experienced meditators and put them into the tube or used EEG to scan them. You’ll see that if you train or practice meditation it requires a lot of cognitive control, engages your attention systems because you have to repress mind wandering, you have to learn to focus, or you have to learn to widen your focus of intention but not let intrusions come. You need certain centers in the brain to do that, and they light up when you do this practice. There’s also evidence that certain cortical structures increase in thickness, namely those that are part of the attention network.

What do you think of rebooting the plasticity in adult brains to behave more like children’s brains? Like, chemically?

That’s what everybody would hope for, especially after injury.

What about technology improving to allow us to observe our own brains more regularly and in more detail? Like fMRI machines in our phones or something? Do you think spatiotemporal resolution will improve so much in the near future?

Well, you can’t carry around an fMRI machine in your pocket, you need a 3-Tesla magnet. You can do EG, very lousy spatio-resolution, because you have all this volume production. You can plant electro chips. You do this with para- lyzed people for them to control a robot arm, for example. I am more on the skep- tical side.

Why?

First of all, we have not understood the essential principles of the brain. Silicon Valley people produce these good-looking machines and neural networks, and they have fantastic performances in classification, but that’s it. Playing Go is nothing more than that. You just have to learn from examples. If you have enough time and enough speed, you iterate these trial and error things until you get a stra- tegy that’s super good. So they outper- form us on particular tasks. My phone outperforms me when I do a numerical calculation. Let them do it. It’s fine. Nice servants. I come from a time when I still used the ruler to calculate logarithmic numbers or tables.

So these tasks are abstractions of biological processes?

Abstractions, approximations, guess work. We don’t really understand how the cerebral cortex does what it does – and it only takes 30 watts of energy. Compare that with what computers use, it’s like a city in order to do the calculations that we can do in our heads. We have much to learn from it. They will have to learn from us as soon as we understand more and then try to implement those principles in... probably not silicon. My guess is that it will have to be another substrate, because much of this stuff is analog computing, and this you can’t do well in silicon. So far they have no technical implementation of a clever learning rule. It all has to be calculated, embodied in a chip. So I’m very relaxed. And I know I’m in good company, because everyone who doesn’t make big money with machines but who instead try to really get at the essence of what generative computing means, they share my skepticism.

We’ve encountered a lot of optimism in tech. Do you think they’ll catch up to your skepticism?

Certain computer people, those who really made the advances on the theoretical level before all the limitations came, now detect and realize that they get the same problems we have: it’s the binding problem, the question of how you represent message relations, how you get a representation of a leaf on a branch from a tree in an environment. You have these many brackets, and in language construction you have the same thing. The way Google Translate does it is it compares the world literature in the ori- ginal language in the input with world literature in translations in the output. And they match it until it fits, but this is not how we do it. We try to get the meaning, we search the right vocabulary, it’s a completely different science. And we call these functions generative functions. These machines cannot do it because they lack essential features of organization that we have. But unfortunately these are features that engineers hate. They hate the recurrent network.

Why?

Because it’s not controllable. You cannot analytically analyze it, not mathematically either because it’s too complex. It’s too non-linear so you can’t really predict what it’s going to do. So it has this runaway kinetics that must be very well controlled or else it gets epileptic or it dies out. And all these problems make them find other solutions.

You could argue that airplanes don’t flap their wings like birds.

But this is aerodynamics. The cognitive principles used by the brain, in my opinion, are still in some respects radically different from those used nowadays in supercomputers. It will take quite some time until we have done our job, and we can build little machines that only consume 30 watts, and start to behave a little bit like a fly. If you look at a mos- quito, and the intelligence of this mosquito – I’m sure you’ve tried to catch one at night – you start to admire these little machines: there’s nothing in the artificial world that can approximate this in terms of energy efficiency and cuteness.

What are the challenges in the next ten years in neuroscience?

Cope with the dynamics and the complexity. We now have the tools, and this is really new in the field. Until we could look at more than one node of a network at the same time, people used to observe one node in different stages of the brain, one after the other. This precludes you from seeing relational constructs. You cannot clap with one hand. As soon as you start with this, you see relations, and you start to see what looks like noise, because A is always doing this and B is doing something else. As soon as you see that these two things are related, it’s no longer noise. The more of these nodes you record simultaneously, the more you see that everything is coordinated with everything in a very subtle way.

So if you look at a single place it looks like noise, but if you look at many places it looks like a pattern?

We can finally do this. With modern technologies, optimal recording, we can look at thousands of nodes at the same time. We get this extremely high-dimensional data, you can’t see anything when you look at it, it’s just dots and curves – you can’t make any sense out of it. So you need machines in order to detect patterns in there – machine learning – and you need mathematics to cope with these complex, high-dimensional vectors. It’s not only vectors, it’s trajectories, and the trajectory of vectors, because activity changes in time all the time. It must, and only because it does do we have a concept of time flowing. If it always stayed the same, time wouldn’t move.

It’s like we need new mathematics.

Yes. We need much more conceptual work to make sense out of the data. We can collect it much better than interpret it. New technologies have opened the field up. We are able to record from thousands of neurons at the same time, we have anatomical methods to see the whole network. You could eventually really trace it, but that doesn’t really help you. What you see is complexity and very high-dimension dynamics. And somehow this goes well together. A complex system will develop such dynamics. The real problem now is what to do with all these facts – how to put them together, what sort of concepts do they develop, how to test them, how to make good predictions for further research, because obviously there’s no point in just collecting whatever you get, as we said initially. You never know if what you have is a side effect not worth pursuing or if it’s the real thing. Before you have a concept you don’t know.

It’s like the more we are able to observe the more we know how little we know.

This is exactly what my feeling is. 20 years ago, I thought I had understood more than I now know I understand today. There is a lot of progress but the insight into not knowing has grown more rapidly than the insight into knowing has.

What about neurological diseases? Do you have any insights into slowing the ageing brain?

There are different aspects. Obviously with degenerative diseases we get more and more a handle on the mechanisms, as well as the genetic causes. Therapy is a big problem. It’s not easy to interfere with these processes. We know roughly what’s going on but we are not yet able to stop it. That may change rapidly with technology, since we can really hunt down genes and manipulate gene expression, but we aren’t there yet. But I do think that we will have a cure for certain degenerative diseases, whether it’s Alzheimer’s I’m not sure. Maybe Parkinson’s. ALS is about to be solvable – at least in the near future.

And what about psychiatry?

It’s very different, because there we don’t understand what the problem is, where it resides. All we know now – and we have a number of conferences on it, the Ernst Strüngmann Conferences, which used to be in Dahlem. We had three or four on psychiatric diseases. The bottom line is the taxonomy, the diagnosis, is very coarse. What we call schizophrenia probably has a very different result, a very different mechanism to something else we call schizophrenia. It’s probably very different diseases, and the same with autism and so forth. So we need a better classification and taxonomy of it before we can do systematic research. We have certain hyimpotheses of what’s going wrong but if you look at all of them, they are not coherent yet. This is partly a reflection that we don’t understand very basic principles of cortical functions sup- porting higher cognitive functions.

Does that also mean that there’s no pro- gress pharmacologically?

Very little. All the drugs that we have nowadays were serendipitously discovered 50 years ago, with added modifications to alleviate certain side effects. There’s no new principle so far. Lithium for depression. So the field is failing, and the field is searching for solutions, and the field doesn’t quite know where they will come from. It’s a big problem. We are helpless here.

What does lithium do for depression? I just watched Homeland recently, and Carrie’s prescribed lithium –

It acts like sodium in the brain, in terms of binding, but it works in some patients because it changes excitability levels. But we haven’t really come to grips with it. Deep brain stimulation has also been developed.

For Parkinson’s?

We don’t know how it works exactly for Parkinson’s. It’s been examined in animal experimentation, and there’s a good concept behind it, and people have realized that when they got it wrong, when they stimulated places that they didn’t want to stimulate, that it had effects on mood. So there was this revival of psychosurgery, which we had already condemned 50 years ago.

Is it bad for the brain?

Stimulation is thought to be reversible, but I doubt it, because if you stimulate the brain over weeks and weeks, it must change something. It’s an active field, trial and error, ethically questionable sometimes, because these interventions, unlike prescribing a drug, are not sub- ject to the same ethical criteria as drug development.

So the FDA is something we should hold on to?

They require endless trials, double-blind and so forth, before treatments are approved. With deep brain stimulation it is enough, because it is a method, if the patient and the psychiatrist agree that they should have an intervention. If they find a neurosurgeon to do it, they can do it. They don’t have to ask an ethical committee and so forth. And of course money is involved. It started in patients that are so-called helpless, who can’t be helped pharmacologically. So desperate cases, who consent because they see it as the last resort. And if you look at what they do, they try here and they try there, and stimulate here and stimulate there. I was directing the Academy of Sciences for a while, asked to analyze the situation, stop it, and do what you have to do on ethical committees. You have to be there for a longterm examination of the development of these patients, follow them for a long time, and do it in a systematic way, and publish, and also publish negative findings. I hope that this will stop this hazardous, aleatoric playing around with brains.

We were in touch with DARPA-associated institutions – the Lawrence Livermore Institute in San Francisco. They’re implanting these chips – electronic devices – in the brain, with the hope of being able to sort of control them remotely. It’s not there yet, but proof of concept is. But for Parkinson’s it seems to work.

There’s something to it. I can imagine that certain forms of major depression can be treated that way. By stimulating the reward centers in the brain. But so far there is no canonical recipe.
We did transcranial magnetic stimulation.

Ah yeah. We have these machines here.

The idea is that early artforms made by prehistoric humans – from the north and the south – resemble each other because trance states activate the visual cortex in the same way that TMS does. It sounds dangerous when you talk about it.

No, TMS is not dangerous. Maybe it can trigger epilepsy.

He just triggered it a little above the neck.

You see phosphenes. You’d have to bring it higher up in the cortical area, and all of a sudden you’d see faces appearing. Imagery. The question is why do we draw stick figures all the time? This might be a genetic imprint. Because the body scheme is so similar in all mammals. A head, a trunk, and four paws. Either you could take the stance that it’s a very high degree of abstraction, or it is the most primitive representation of a mammal. That’s always the discussion, right?

What do you think?

I think it’s both.

What do you think about virtual reality?

Ah, great opportunity for art. I’ve been to several symposia recently on the chances of using virtual reality and augmented reality to embed the observer much more in the piece of art. Because it can really absorb you completely, which looking at a painting can’t as much. I know Daniel Birnbaum, the former director of the Städelschule.

Yeah, I studied there. He taught a philosophy seminar.

Ah, I know him well. He is moving now from the modern art museum in Stockholm to a company that does virtual reality, because he wants to make this technique available to artists. It’s certainly something that one should keep in mind. Cinema started to outperform theater to some extent. This will certainly replace the current video monitors in exhibitions.

In terms of simulating experiences, it can also work much better with emotions like empathy.

Yeah, yeah. Because you can fool the brain if you simulate the sensory evidence. You can also take a flight simulator at the airport. They have all the noise and the vibrations.
There you can really embed it. I saw these pilots sweating. There were sitting in a simulator, and they really felt they had to do it right. So you forget very quickly that you are in a simulator.

We saw Star Wars in 4DX. The seat was rattling, and there was a plastic tube that tickled your legs, and water was squirted in your face. I put my jacket on, it was freezing.

VR becomes reality again. It capitalizes on the knowledge that the brain has about the world. You give it a few things to eat and it will reconstruct the rest.

© Yngve Holen 2019. ETOPS IIII, edited by Yngve Holen and Mathew Evans.

What do you want to know?

How did you end up here?

I was invited to, uh... I was given the job 36 years ago. And I’ve been working with INPA ever since.

We spoke with your colleague, Charles.

Uh huh.

And we’re in touch with Susanna. Something we had discussed with them are the challenges of biodiversity. How to manage these unknown quantities, put value on something we don’t know much about. How do you fund biodiversity management?

Well, from my end we have almost no money. Brazil’s in a crisis. But before Brazil was in a crisis we had very little money. We did a calculation and it came out to be something like 1 cent per hectare per year. It’s almost 1/100th of a cent per hectare per year for the most of the Amazon. People have trouble because they can’t comprehend how big the Amazon is, and just how few researches there are, and little access there is to it. People often think of a 1-million-dollar project, which is an awful lot of money for most studies people do in the US or Europe, where you’re talking about a couple of square kilometers, and there’s easy access to get there by road or trail. But here, when you spread that over 7 million square kilometers, it’s very little. That’s our biggest problem—the scale of the area. Things can happen that you don’t know happen, because you’re not there. And local people don’t have the education. They have the laws, lots of things to back them up, but they don’t know their rights. So they tend to get walked over by big business and corporations in the Amazon.

And it’s hard to control that business, because of limited monitoring possibilities. You don’t know who accesses what, and what they do with that.

It’s still easy... Well, it’s not easy. There are a lot of people in the Amazon. So if we could get that work force organized, then they would do a lot of protecting themselves. But at the moment they aren’t thinking about that. They’re thinking about high technology themes. Thinking about going in and finding that leaf that will cure cancer. And very few people are thinking about the Amazon as a productive system, and that there are lots of people living off it. And other things like world climate are so popular.

What many people across the world consider the project of saving the Amazon has actually very little to do with the people who live here.

People think that there’s almost nobody in the Amazon, but there are millions of people here. And the sorts of agriculture they had done up until now had sustained them, and had sustained the ecosystem. If you replace that with large-scale monoculture and the amounts that the ecosystem has to process, you lose the people, and you usually don’t make a lot of money either. There are a whole lot of perks and strange business deals that go on. A few people make a lot of money. But for the area, there’s very little production.

There’s all these factories here in Manaus. We’ve been here for about 5 days. Electronics factories, automobile factories. We see them everywhere. How do you look at Manaus as a city that’s inte- grated in the rainforest, that works with it?

In a way it does. Manaus generates a lot of money. And that money can be used to sustain people and for education and all sorts of things. And Manaus effects a very small area in terms of the forest, because it makes little demands on the forest. It’s sort of a little enclave. There’s virtually no hinterland around Manaus. No agricultural production. Most things are flown in. The Free Zone—the original idea was put in by the military government. In order to attract a lot of people to the Amazon, they would cut down the trees. And the military also thought that communists wouldn’t be able to hide in the grass then either, so they would solve their problems. They wanted to get rid of the jungle. It didn’t work that way. But what it basically did was make an enclave that generates a lot of money, which is good in general. The State of Amazonas is one of the most preserved states within the Amazon. But it’s also one of the richest states. That’s because it’s living on a Free Zone. And so there’s all sorts of opportunities for high technology, and there’s also more opportunities for education and healthcare. Other states like Pará and Rondônia and Acre, they don’t have those benefits, so people try to make money by cutting down forest and putting in cows, soybean, and other things. I’m not against the Free Zone for very large reasons.

The Amazon stretches across 6 countries. And those countries all don’t have the same political or ecological agenda. There are also European and Asian countries coming in, interested in the hydropower possibilities, for example. What kind of environmental scope does scientific research have in and of itself with what you do?

One of our biggest projects at the moment is to do an environmental evaluation of all the sites where hydroelectric dams are planned. At the moment, the government requires an environmental impact statement. The problem is that they’re already putting in the dam when they’re doing the environmental impact statement. And it costs lots of money to take away the dam. So what we have to do is go in when they’re planning the dam, because they plan these dams 20 or 30 years before. So if we can go in and evaluate the biodiversity in one of these places, we might be able to determine the potential—the money potential—for the area, or the potential loss if the dam is built. And then the idea is to try to optimize, to get as much energy for the least loss of biodiversity. We have to do the surveys now, and that’s what we can’t get money to do. If we invest now, we save money later. But there are a lot of vested interests in not planning, and there are even vested interests by the biologists, because a lot of biologists get rich doing these statements that are used for nothing. So one of our biggest projects at the moment is to do the integration of environmental planning and biodiversity. But another big problem is in countries that supply the water to the Amazon. They’re putting in an enormous number of hydroelectric dams, and we have very little influence there. Though we are starting some training programs with countries like Ecuador, so we can influence what goes on in those other countries. But it’s much harder.

Yngve’s from Norway. And we know that Norway invested 100 million euros or something into saving the rainforest.

We also heard quite a lot of joking around about what’s happening with that money. Charles said that he went down to Brasilia to this meeting where there were... People were brainstorming about how to use the money. And he said if you’re going to save the rainforest, why don’t you just build a fence around it, with guards? Why invest in a bio pepper seed?

Why invest heavily in the extraction economy? Why not just put a fence up?

I talked to the Norway people, but they were really interested in investing in carbon. And carbon trading. And that’s a good example of the size of the problem we have. So they have all the money, but they may spread it around so thinly that they end up not being able to do anything. Because what you have to do is integrated planning for the whole of the basin. And they end up forking out bits of money here and there, to this and that, and although it might be very useful from the point of view of a normal academic program, it’s not... What we need is a lot of planning over the whole of the basin. And it has to be integrated. 

And what about this fence?

I think that they’re not against... They want to work with the people, give them other opportunities to make money. It’s just, as we don’t have a good plan for the whole of the Amazon, we don’t know where it’s best to invest. So people have been putting lots of money into the Amazon for quite a long time now, in terms of billions of dollars. But it just seems to disappear. It’s not going into a system that builds for the long-term. So I don’t know how this Norway money will finally come out, but we certainly haven’t seen any of it.

There seems to be a lot of unknown aspects of the Amazon.

Yeah, that’s the thing. There’s a movement within the biodiversity section of the Ministry of Science and Technology that’s trying to be more efficient in that way. But that movement came just at the time when Brazil went into a crisis. When we had a little bit of money, there was no planning, but now that there’s no money, we’re get- ting around to the planning.

How do the Amazonian people feel about monoculture?

You see, there’s very little organization and there’s no communication. So people then see it. These things just eat away the Amazon, and those local ideas and knowledge. These people are all very poor, need money, and they don’t have very much employment. So cutting down the forest looks good to them for a while. And then they find that when monoculture goes in, there’s no more forest. They can’t live there anymore, so they might go to the periphery of the city and become city dwellers. People don’t see it happening. It just sort of creeps up on them.

So there’s no real economic alternative to chopping down the rainforest?

It’s not going to happen, because we’ve got a lot of poor people out there, and they’re worried about the health of their kids and better conditions. And the only way they see they can do that is to have tire roads and cell phones. And they don’t see what’s happening. It just keeps nibbling away. The only place that you sort of have grassroots resistance is way back when they tried to go into the rubber cutters and take them over, and cutters refused people from going in.

Chico Mendes, in the 80s.

Yeah, but that’s 30 years ago. That sort of thing is not happening now at all really.

We’ve been hearing that they’re trying to create more economy through extraction products. Brazil nut, açai.

Well that was... The idea then and now is that the people would have standing forest products. And basically they’ve never had the support. They’ve never had the government doing those tests in those production systems. Because it’s very different to invest in a lot of very small-scale farms, which produce a small amount each. You have to have an organization to get that going. Each person working individually doesn’t work. You could say the same thing about Europe. You could say the small farms in Europe, they’re trying to keep them going in Europe, but in order to do that they have to have some sort of government intervention. And here it’s even more difficult. And so the Sustainable Development Reserve in the other reserves, the people don’t have the lifestyle, the basic education and health requirements, and it’s really hard to make that work in the long term.

Aesthetic industries are never going to be efficient enough to justify the fate of the Amazon. Basically, the Amazon does provide all of the water to all the agricultural growing regions of the southern part of Brazil, and there’s all these carbon things. The world wants the Amazon standing, so the Brazilian government wants the Amazon standing. But they don’t want to invest the very small amount that’s necessary to make the lives of the local people better. So the local people don’t see any alternatives except cutting down the forest. So what’s really needed is strong government planning to maintain the forest. The state of Acre even subsidizes the rubber, because it’s better to subsidize the rubber than have people cut down the forest. But we don’t have that on a larger scale. We need it, but we’re not going to get it. Instead of people saying, we’re going to open up new roads into the Amazon, make new places to get more money... What we need is for the government to consolidate the places where there are people, consolidate the infrastructure, consolidate the health and education. And that wouldn’t take that much money, instead of wasting money making new roads. You can see the same sort of thing in the south of Brazil. 40 percent of Brazil’s agricultural production is just wasted. It never actually gets to the table, because of the very bad infrastructure. They’re always trying to start something new instead of investing in consolidating infrastructure that’s already there.

Maybe some kind of longterm partnership with an outside investor? One of the energy partnerships? Or is that asking too much?

There is big hydroelectric potential in the Amazon, and there’s no way that people aren’t going to use it. It’s just to use it sensibly. They put in Belo Monte, but in a few years there won’t be enough water to run it. And so they’re going to say, we’ve invested all these billions, and the problem is the Indians who won’t let us put the other dams upstream. They won’t let the water flow into it. And then they’ll want to go into the Kayapo lands and destroy those. And this frontier mentality... Instead of saying yes, you’re not going to be able to preserve the whole Amazon. People are going to want it, and they’re going want to use it in new ways. But we should go carefully, consolidating. And part of that consolidation is the products, the small agriculture, the standing crops, the tree crops. And putting it in the system so they function, instead of trying to expand. We have to get rid of the frontier mentality. That’s the problem.

So what is the Amazon if it’s not a fron- tier. If we think about it globally?

Firstly it’s a mitigation system. It stores 1/4 of the total carbon.

Hence the carbon trading.

Well, yeah. To not get too complicated... We’ve been discussing dams. The Amazon is a giant water pump. Basically, the northeastern trade winds, as they go past, they collect water vapor. And that brings water to Latin America. The rain falls in Uruguay, Paraguay, the plains of the basin. Argentina. Mato Grosso. All these areas that are the grainery of Latin America. These agricultural economies are worth something like 1 trillion dollars. Of course getting rain to those regions is necessary for those economies to function. Of course no one pays Amazonas for providing that function. And that of course is a problem we have globally of not recognizing public goods and services.

It’s also the North-South issue.

Globally, yes.

Nice caiman head. And that crocodile over there. Where does your fascination come from? Are you from Australia?

Yes.

Because we went to the floating forest, near Tefé. And we saw caimans for the first time and they were terrifying crea- tures. What’s attractive about them?

Everyone’s attracted to seemingly terrifying animals. Tigers, elephants. There are all sorts of things that are terrifying. But I don’t know. I had an opportunity to work with them when I was in Australia. Caimans—like this one—are pussycats compared to crocodylidae. If you were up by Tefé, you saw these caimans that were up there floating around the lodge. But if they were crocodiles, they would eat people. It’s very different. So I just... There was a time when it seemed like it would be a good opportunity for local people to hunt caiman. They’ve always hunted caiman. They’ve always eaten caiman. Not so much in the State of Amazonas, but they export it to the State of Pará. For a long time they sold the skin to the luxury industry, and it seemed like for a long time that it would be a good industry for the people. But they’ve run into all sorts of problems, mainly from the agricultural industry, which wants us to kill them like cows. Take them to a slaughter house, and hang them up. But that doesn’t work for a wild animal. Although it does work for some wild animals, but not for these. They just become inedible if you do that. It’s the only country in the world that requires that. All the other countries in the world—like Australia, the US—where they hunt crocodilians, they treat them like fish. So there’s a big opportunity for the local people to have... to make money out of these cai- mans, but we’re not making much use of it yet.

You can make more money from the skin than the meat probably.

Both. The skin has a reasonable value, but it’s the meat that makes it worthwhile. If you can’t sell the meat, it’s not worthwhile to hunt. It’s also ethical, to go out there and kill a big animal like that. Just taking the skin and throwing the meat in the water.

They have an amphibious style.

Well, yeah, in the sense that it lives in the water and it nests on land. It can walk on land. They’re not amphibians. They’re pretty interesting animals. The can do all sorts of things. They have the most complex vertebrate heart. They can do all sorts of things that birds and mammals can do, and they can do all sorts of things that amphibians can do. They can swim like a fish, gallop like a horse. Their eyesight is much better than ours. They see well at night, in black and white. They also can see more colors than us. They have color vision like birds. So a crocodile looking through our eyes would think that he’s color blind. They do all sorts of amazing things. They’re special creatures. And they’ve been around a long time. They used to feed on dinosaurs, and they’re still feeding on us, and they’ll still be feeding on whatever dominant group there is on the earth 50 million years from now.

I read somewhere that the crocodile was around when the mammal was still extremely small. It wasn’t evolved into a large order.

Mammals were pretty small back then, skidding around. They didn’t take off until the dinosaurs died out. Though all of the archazoes died out at the end of the Jurassic Period, except for the crocodilians and the birds. And the birds are dinosaurs. They’re just dinosaurs. But all of the other dinosaurs died out. Only the crocs and birds survived.

Is there any explanation?

Nobody really knows. There are a whole lot of freshwater things that survived. Possibly because things that live in fresh water, they can feed on dead things that fall into the water. They think that basically a meteorite hit the earth, and cut out all the sunlight for years. So there were no green plants producing anything. And so the animals eating the green plants died out. How the birds did it, with their high metabolic rate, nobody knows. But perhaps they were able to migrate to find the little bits of food that were left.

Felipe told us to be in touch with you. Do you have any relationship to his restaurant? I know that INPA and the restaurant are trying to work together. Do you have a food passion?

I coordinate a very large project called CENBAM. It’s the National Institute for Science Technology and Innovation for Amazonian Biodiversity. And one of our researches is Noemia Ishikawa. And Noemia studies the fungi. And she’s interested in that. She’s descended from Japanese immigrants. And she had a lot to do with—I think it was her grandfather who started growing shiitake in Brazil. And so when she came here, she’s been trying to promote the use of the Lentinus strigo- sus species. She just spent some time in Roraima, with the Indians up there. The Yanomami. And they have dozens and dozens of species of mushrooms that they eat. So it looks like it could be an industry. Either an extracting industry for the Indians, or an opportunity to grow them. Noemia’s been working with the people who have the brazil nut plantations, and they have to cut off the excess branches from the brazil nuts, and they use those excess branches to grow shiitake. With all of that, it’s no good having production unless you can sell it. So people like Felipe have been helping out by looking at what the economic potential of those things could be, because you need a whole production chain. It’s not enough to just go there, and say, someone can grow it, someone can harvest it, and someone can sell it. You have to make sure you have a whole system that will keep going and be consistent. So that’s why we work with Felipe.

What kinds of technologies are used to map biodiversity?

Well, there’s been some advancement. And people of course are looking into ways to better map species. Mosquitos, for example. Which are very dangerous, and very expensive to get rid of.

Didn’t the Americans kill them off with gasoline when they were building the Panama Canal? Just drenched the whole landscape with gas.

They’re still trying to commit specicide on around 30 different kinds of mosqui- tos worldwide. But there’s thousands of different species of mosquitoes, and if we get rid of a few, the landscapes will hiccup, and who knows if something better or worse will then emerge. We’re still a long way away from knowing the extent and complex behavior of these ecosystems.

What about smartphones? Bird apps?

Those are being developed too. [Laughs]

Is it like sleep? With all these apps we can now test sleep patterns more generally.

Nature doesn’t use the same apps as us. People are really interested in these things. Things like the food and the small industries and what not. We do work with that. But it’s not going to make a big difference over the whole of the Amazon. The mosquito technology is useful immediately, but that’s a whole other story. Our technology is about environmental impact statements, and how you can do biodiversity surveying, and put it into a system where you can do conservation planning. So the system that we developed is on biodiversity monitoring. That system, the repel system, is used as a basic-impact evaluation system for dams. It’s used by Obama. We work both with the federal government and the private sector. We do a lot of training. We’re developing that. Each one of these little industries that comes up is really interesting, and everyone likes to do it. If you can stand there and hold out: I’ve developed this tomato. It’s something that’s palpable. But it’s harder to get people interested in long- term planning. Because nobody sees it. If you plan well, everyone expects what you should have done. If you plan badly, everyone complains. If you do it well, everyone thinks it’s normal. That’s our main line of attack— get people planning between the various sectors, like mining, the energy sector, the transport sector. So that we can integrate that with biodiversity.

How is the political sphere responding to this?

Surprisingly well, but slowly. There’s not much money. That’s the whole problem. As we were starting to get there, the money dried up. There’s often resistance from places you wouldn’t expect. Like from biologists, who have been used to doing their own little research and not having to interact with anybody, and they publish it in some journal, and it has very little impact. So it’s really changing culture. And scientific culture is very entrenched. We actually have... It’s not more difficult to work with politicians than it is to work with the biologists.

We spoke with a chef who has this restaurant in São Paulo. Fancy restaurant. It also has this social-ecological branch, called ATA. And they help the Baniwa produce this really delicious pepper in the Northwest.

© Yngve Holen 2015. ETOPS III, edited by Yngve Holen and Mathew Evans.