When I look through the window of my flat in Warsaw I see a shop. The shop is called Organic; it is full of overpriced linen seeds, nuts and, most interestingly, different kinds of salt. I am not sure in what sense salt can be organic, but I think I know what are the benefits of calling it that way. Organic means biological. Biological, loosely, means natural. And natural means good.
Organic salt is, thereafter, good salt.
Organic salt is just one of the examples of the infamous appeal to nature. For some reason, at least in Western culture, things that are thought to be natural seem automatically good; better than unnatural ones. That belief is the root of natural medicine, some non-vege people arguments (but humans are carnivores!) and the existence of the Organic shop on the other side of my street.
Things that are natural are also true – that’s why many philosophers believed that only a man in the wild, primitive state is a true human. And that’s why many evolutionary anthropologists study people in tribes isolated from civilisation – because it might tell us something more about the true human nature, one not tainted with the culture that is supposed to run against many of our instincts.
This is also, in my opinion, one of the reasons why people in behavioral neuroscience, including myself, are nowadays becoming fascinated by ethological relevance.
Development of new methods – like miniscopes (for people outside of the field – these are miniaturized microscopes that allow you to image brain activity in a freely-moving animal) – and the constant development and improvement of others (e.g. ephys); methods that are very often open-source and thus cheap, are definitely the main motor of changes in the field. After decades of doing behavioral studies on immobilized, head-fixed animals people are developing paradigms in which the animal is free to move and explore the environment much more eagerly.
But this fascination goes further than just letting the animal go around with a miniature microscope on its head. More and more I hear neuroscientists claim that they are trying to make their paradigms more natural – or, as I said above, ethologically relevant.
Ethological relevance has many faces. It can be as simple a using stimulus that is supposed to mimic a natural one – e.g. a looming black circle that imitates an approaching aerial predator. It evokes an instant escape response in mice; and, even intuitively, reproduces something that probably occurs frequently in the life of wild mice, especially if you compare it to electric shocks that could at best imitate an attack by a very weak electric eel (see figure above).
It can be something more – instead of using a natural stimulus, you can try to create a whole natural environment. You want to study social behaviors? A few years ago you would put a mouse under the mesh cup and let the other one interact with it. If it does, it is social, if it prefers a non-social object (or an empty cup), it is probably not – and maybe even autistic!
Seems silly? Well, it probably is; but now you can do something much fancier – like putting mice for a few weeks in a large, automated cage composed of many chambers connected by tunnels imitating burrows, and study how they interact with each other without even a need of touching them, just like in a device that was developed in our lab.
Even those who are, for methodological reasons, forced to head-fix their animals are trying their best to make their studies a little bit closer to Mother Nature. Andrew Fink and Carl Shoonover developed a virtual burrow. The idea is simple: mice live in burrows and feel safe inside them. Why not put a head – fixed mouse inside of a burrow and do it in a way that will allow the animal to go outside to explore or hide when needed? It can be used to study curiosity, anxiety and who knows what else; and you can still do your two-photon imaging.
As you may have guessed from my title, this essay will be a critique of the ethologically relevant approach. But just to be clear, I want to stress: it certainly has many benefits.
First, there is the question of stress: in the two last examples animals are probably much less stressed than in traditional paradigms, which is always a good thing.
But there is more. Some people argue that paradigms that use unnatural stimuli may evoke unnatural responses in the brain. Mice are not prepared by the evolution to the presentation of white and black stripes on the screen while being immobilized. Their brains might not work as they would if the mice were watching something they usually see in their short lives – and it might be the case (similar problems can be found in fear studies, look here).
Furthermore, behaviors that are instinctive should be possible to see most clearly in natural circumstances in which they evolved. And if many instinctive behaviors are conserved from human to mice (if that order feels weird, read that wonderful paper), you might learn a lot about human behavior studying instinctive behaviors of mice in naturalistic paradigms. And people do that, obviously; all those studies on feeding, fleeing, fighting and mating are here also to help us understand our basic instincts.
These arguments seem to me to be reducible to one simple statement: that by studying more naturalistic behaviors we have bigger chances to study something real. But well, it might not be always true; and here we come to the critique.
We have a strong tendency to believe in a fixed nature of things. Ernst Mayr once claimed that people came up with an idea of evolution so late because of Platonic essentialism – a belief that things have their true essence. As Richard Dawkins put it, if you treat all flesh-and-blood rabbits as imperfect approximations to an ideal Platonic rabbit, it won’t occur to you that rabbits might have evolved from a non-rabbit ancestor, and might evolve into a non-rabbit descendant.
In my opinion, we show the same essentialistic thinking when we describe animal behavior. There is a defined set of behaviors, the natural repertoire, that is specific to mice. Anything outside of this set – like pulling a lever – is un-natural and artificial.; it’s untrue.
In this framework, animals are viewed a little bit like robots, endowed by a designer – in this case, natural selection – with some well-specified functions. If your robot was designed to wash your clothes and make you healthy dishes full of organic salt and linen seeds, you can expect that it will do those things well. You can try to slightly modify it to do some other stuff – like trolling flat Earth supporters on Twitter – but then you cannot expect it do perform perfectly. If it does it, you are lucky, if not – well, life. The most probable result, though, will be some variable, mediocre, unstable performance.
But animals are not robots. They live in an environment that changes, and usually changes quickly. Our beloved laboratory rodents are descendants of animals that adapted to a completely new environment of human settlements, that were itself changing fast. Obviously those animals were evolving; instincts were moulded by the new pressures. But animals also learn how to solve new problems and they are doing it well – just look on the corvids using cars to smash nuts, or blue tits learning how to open milk bottles, or a whole book of such examples. The nature of (many) animals is the lack of a fixed nature.
Animal’s brains are prepared to do new things; they are also prepared to learn how to pull a lever. You can even make a very controversial claim – if an animal is able to learn a task, it is able to learn this task, and it’s using its brain to do that. It means that you can use this task to study how the brain learns – even if at the end of the day you will only know how the brain learns very weird things. Learning very weird things, though, is the stuff that the human brain does most of the time nowadays.
Moreover, the big problem with studying natural behaviors of mice or rats is that we know relatively little about their natural behavior. We have a somewhat cliche image of a wild mouse: it’s small, it lives in burrows, it’s afraid of predators and well, that’s almost all we can say. There are very nice, but quite old books: Mice all over by Crowcroft and, for rats, Rat – a study in behavior by Barnett; we have also recently put some effort to bring together what is known about social behaviors of mice and rats, but we need much more studies on the rodent behavior in the wild to do a really naturalistic neuroscience.
As we discuss in the aforementioned paper, laboratory animals that we use are domesticated animals. And it poses yet another problem. Our lab animals differ in behavior from their wild counterparts. They are less aggressive; their sexual behavior, a supposed pinnacle of instinctiveness, is altered; the way they learn or flee might be also different – natural behaviors of a wild mouse might not be at all natural to a laboratory mouse. You will not learn much about wolf hunting behavior by observing your chihuahua, which is basically a wolf strain, the C57 of Canis lupus*.
Don’t get me wrong: I am a great fan of ethologically relevant studies; moreover, I am trying to do them as well. What I am arguing against is making a dogma of it. I have recently talked with a person who claimed that people doing head-fixed task on rodents are reductionist. They want to reduce en extremely complex phenomenon of learning to the passive process of watching sequences of colorful dots for thousands of times by a poor, immobile mouse.
Nicole C. Nelson in her wonderful book made an interesting point of calling this approach reproductionistic instead of reductionistic. She studied scientists working on mice models of alcohol addiction. According to her observations, they are fully aware of the complexity of the human condition; and they do not want to reduce the socio-psycho-bio-who-know-what-else phenomenon of alcohol addiction to C57 mice drinking ethanol from the bottle. Instead, they want to reproduce some aspects of this phenomenon in mice just to make it open to a scientific investigations. Mouse models of alcoholism have innumerable problems and they will never allow us to study how poverty can make us addicted to alcohol. But they can be to some extent helpful when put in a bigger context of discoveries from other fields.
Those of us who work with immobilized mice watching dots do not usually claim that they will solve the problem of perception. They just reproducing some aspects of perception to make it tractable.
Gustav Flaubert composed a beautiful list of slogans and cliches popular among people of his time. According to it, THE PRESENT AGE should be always denounced vigorously; when somebody talks about ANIMALS, usually mentions that some of them are more intelligent than humans; if you mention PAGANINI, you are supposed to say that his fingers were enormously long.
Flaubert despised cliches. According to him, they are automatic expressions that we say when our thought is lazy; they could have been original or creative when they were said for the first time, but now are dead, they are meaningless. We can have impression that we said something funny, interesting or deep, but it is just a zombie of thought.
An idea of doing ethologically relevant neuroscience, although not new, is still fresh and viable. I am a little bit afraid that by not thinking deeper about what does it mean to be ethologically relevant, when it is advantageous and when not, by designing studies that are naturalistic only superficially, and by dismissing too easily studies that are not, we will turn it into a cliche. Naturalistic could become a mindless label that we will use to easily judge the study. Absolutely absurd paradigm could be then perceived as interesting just because somebody used hiperrealistic robot model of a buzzard instead of an electric shock.
Let’s not go there, let’s think; let’s keep ethological relevance alive as long as possible.
* That’s an obvious overexaggeration; C57 was developed in 1921; dog strains have much longer history. But see Byelayev research on siberian foxes – in this case their behavior and morphology changed very quickly under domestication, so you do not really need many generations to change an animal (but there is some recent criticism of this study, a criticism that made me less eager to tell the story of doglike foxes as my favourite lunch anegdote – look here)