California Institute of Technology neurobiologist David J. Anderson is at the vanguard of studying the neurobiological foundations of emotion. He was the key speaker at the Eighth Annual Cell Press-TNQ India Distinguished Lectureship Series. In an interview, he discussed what brain imaging experiments in rodents and flies are telling us about their ‘emotional lives’, whether brain mapping can be used to read minds, and how artificial intelligence could guide future neuroscience research. Excerpts:
In one of your recent brain imaging experiments with mice, you demonstrated that something as basic as a rodent’s ability in telling apart males from females of its own species isn’t necessarily ‘hardwired’. Rather, these appear to be ‘learnt’ abilities. What does that tell us about the role of nature and nurture in shaping our brains?
The more we learn about how the neural circuits that control supposedly innate behaviour are wired in the brain, the more we realise that there is experience-dependent plasticity that sculpts those circuits. Some components of the behaviour are hardwired, but what you direct those behaviours towards isn’t.
An analogy would be language. We are born with the ability to learn language as it is hardwired into our brains by evolution, but which language we speak depends on where we grow up. The behaviours that I study with my research group — aggression and mating behaviour — are, in a sense, hardwired. The animal doesn’t need to be trained in how to do those behaviours, but it needs experience to express them correctly, namely, aggression towards another male competing for a female and expressing mating behaviour towards females.
The latter appears to be the first social behaviour that is expressed. The sexual experience, we learn, is what makes the animal aggressive towards other males. This changes the wiring in their brain — or the part of the brain we study — in a way that we can tell from the brain images alone whether the mice have encountered a male or a female.
What other characteristics can be deduced from brain images alone?
We can deduce the type of behaviour that the animal (a mouse in a cage) is engaging in. When another mouse is brought into the cage, we can say from the pattern of brain activity in a specific region (of the first mouse’s brain) if it’s fighting, ‘sniffing’, or investigating if the new mouse is male. If it’s female, we can predict if it’s sniffing or mounting (the female). This is just by taking the patterns of brain activity, performing a computational analysis and asking whether we can train a computer algorithm to predict the behaviour the animal is engaged in. Now, in addition to the type of social behaviour, there are other features, particularly those that are related to emotion. My colleagues and I have tried to come up with some general properties of emotion states that would be reflected in emotional behaviours and would distinguish them from, say, a few robotic reflex responses. One of these critical features is persistence. So emotion states tend to persist for some time after you are exposed to the stimulus.
You can say that from brain images alone?
That’s what we are looking for. The classic example I use is, if you are walking in the woods and suddenly see a snake. You would jump out of the way but even though you are safe and the snake goes away, your heart continues to beat rapidly, you still feel anxious. We are starting to see signatures of that kind of persistence in certain regions of the brain. Whether that persistence is causing persistent behaviours we don’t know yet, but that’s the kind of thing we are trying to extract from this type of brain imaging technique.
Can such knowledge be applied to people?
From those types of studies, I don’t see an immediate application. Right now, it’s more about getting basic knowledge of how the brain works.
However, there are some particulars that we have learnt along the way that could have some applications to humans. For example, one of the things we discovered in the course of studying social behaviour control centres in the brain is that the neurons that control aggression — and to some extent mating — in males are neurons that express the oestrogen receptor. Now, many people find that surprising because the popular view is that aggression is controlled by testosterone and female behaviours by oestrogen. That, however, is only partly true. Other researchers in my field have shown that many of the effects of testosterone to promote aggression are mediated by its conversion into oestrogen in the brain. There is an enzyme in the brain called aromatase that can convert testosterone into oestrogen. The two molecules are so close to each other — it’s a very simple chemical reaction — and in many species it is known that if you block that enzyme, you can block aggressive behaviour. There are drugs that block aromatase that have been used in humans for decades, but have only been used in females as adjuvant chemotherapy for breast cancer. They’ve never been tested in males to see if they might reduce anger, irritability, or aggressiveness. That’s something we’ve been quite interested in. We’ve been collaborating with some clinicians in the U.S. to see if we could carry out open label trials to see if they can, in low doses, be effective in U.S. veterans in treating post-traumatic stress disorder, where domestic violence and irritability is a significant problem. However, it’s been very difficult to raise even small amounts of money to test this in the U.S. because these drugs are off-patent, available in a generic form and offer no profits for companies in finding a new use for an old drug. That was a surprise — almost a shock — to me.
Another issue is a more significant one. We work on both mice and fruit flies. Mice, being mammals, maybe closer to humans than flies, but they’re still not human. There’s a dismal history of failure in the pharmaceutical industry in translating studies of emotional behaviour of anxiety in rats or mice to humans. Most drugs tested, based on those kinds of studies, have failed in humans. Almost all major American pharmaceutical companies have abandoned the search for new neuropsychiatric drugs. There hasn’t been a fundamentally new neuropsychiatric drug in the last 50 years. We have to think hard about what the reasons for that could be.
The first possibility is, as we said, mice aren’t human, and cancer has been cured in mice many times, and yet, very few of those cures have translated into human ones. We are talking here (in the case of cancer) of a level of cell biology and physiology that is much better understood than (the physical basis of) emotion in the brain.
On the other hand, there are some very exciting advances in cancer treatment, like immunotherapy, and these have resulted out of fundamental research done in mice. That gives one some hope. When it comes to psychiatry, some say you will never learn anything from mice applicable to humans because their brains are too different. For instance, we have a very well developed pre-frontal cortex which is popularly believed to control executive function and exert a sort of top-down inhibitory control on emotion. This was based on evidence from a few lesion cases (results from brain injuries that suppressed or heightened certain emotions and functions). So, they argue that the rodent pre-frontal cortex isn’t as developed as it is in people and so, if psychiatric disorders involve the pre-frontal cortex, then it follows that we need to study nonhuman primates.
There’s a movement, particularly in China and Japan, where regulatory restrictions on working on nonhuman primates are not as great as they are in the U.S. to study brain circuits in rhesus macaques and marmosets. That’s a direction worth following, though I’m not persuaded that all psychiatric disorders are due to dysfunctions of the pre-frontal cortex. There’s plenty of evidence for interaction between the pre-frontal and emotion circuits in mice.
At different points of history, there have been various metaphors to describe the functioning of the brain. In the early 20th century, it was hydraulic view. Then it moved to the brain being seen as a kind of computer. With technologies now to see the neurons in action, is there a different idea of the brain emerging now?
I would say those metaphors are still in active use but it depends on the part of the brain you study. Those who study the cortex tend to think of it as a computer because they are trying to study higher-order cognitive functioning such as decision-making. They think of the limbic system, or the emotional part of the brain, as the hydraulic part of the brain with the view that it doesn’t require the same degree of computational sophistication as the cortex, which has more recently evolved. I think that’s a view flawed in principle as well as from evidence. Our recent data from imaging the hypothalamus shows that many of the properties attributed to the ‘electronic’ brain in the higher cortex, such as experience-dependent plasticity, dynamic coding, etc., are seen in the subcortical, limbic region. That, for many people, is a surprise.
The second misconception is that this [subcortical] part of the brain is the ‘reptilian’ brain and the cortex is more evolved as if the limbic region stopped evolving and it was only the cortex that evolved. That’s flawed because there is such interconnection between those two systems that in humans, both have evolved in parallel. In people, both these systems are highly evolved.
So that means we may have misunderstood or underestimated the complexity of animal brains, fruit fly brains?
Absolutely. In the case of the drosophila (fruit fly), we can say there are regions in the brain evolved to perform certain functions, analogous to that in the human brain. So, for instance, the ‘mushroom body’, which is the structure involved in memory and learning in flies, is functionally similar to, for example, parts of the hippocampus, though it looks nothing like it.
One of the things we’ve discovered in our parallel studies of mice and flies is that in both species — separated by 500 million years of evolution — there are relatively tiny groups of cells that can control both aggressive behaviour and mating behaviour. So in the mouse brain, which has about 100 million neurons, I’m talking about a cluster of about 2,000 cells. In the fly brain, with 10,000 neurons, I’m talking about a cluster of 10 cells. Whether that’s descent from a common ancestor or convergent evolution, we can’t say. But we do find that there are certain motifs in drosophila that are analogous to what is happening in human brains and because there are fewer cells to study in the fly, there are more chances of making links between different cell types, their anatomy, function and behaviour.
