An interview with Pavan Ramdya

You introduce your research by saying that "No two animals behave the same." What does that mean?

When we measure animal behavior – even among animals that are nominally genetically identical – we notice that they actually have strikingly variable behaviors. This is something we also observe in humans: siblings and even twins can have quite different tendencies.

Traditionally, neuroscientists have focused on the “average” behavior of an organism. But now there is a growing community that wants to understand how nervous systems differ, giving rise to individuality.

The reason I find this interesting is because it's one thing that truly distinguishes biological organisms from robots. When we build a robot, we expect it to behave reliably in a specific way. That's actually a positive attribute for a robot. But we will start to expect more flexibility and adaptability from our robots and it will become more and more important to understand how to incorporate some degree of behavioral innovation to our control algorithms – allowing individuals to specialize, and communities to evolve.

What is behavioral innovation?

“Innovation” is a broad term that can mean adaptation by individuals during their lifetime – through learning for example – but also the adaptation by species across evolutionary timescales.

Your field includes “neurogenetics”. Can you tell us what that is?

Neurogenetics means different things to different people. For me, it means studying not only the physiology of neural circuits but also manipulating neuronal gene expression to understand how these changes restructure neuronal activity and behavior.

Can you give us an example of behavioral innovation through neurogenetics?

A common example is the way that sensory systems adapt to environmental cues. For example, animals can evolve olfactory sensors to detect novel odors or to respond differently to an known odor. There has been a great deal of effort to try to understand the olfactory system of Drosophila and scientists have recently identified how changes in specific odor receptor genes allow animals to detect scents in new ways and to change innate behavioral responses.

Moving forward, what's become really interesting is to understand what happens downstream of peripheral sensing. I believe these central circuits will be a critical site underlying behavioral differences across individuals. However, things get a bit more challenging when we begin to target circuits downstream of the sensory periphery.

How can we overcome these challenges?

In the last 10 years, new tools have become available that allow Drosophila researchers to comprehensively and very deeply investigate genetically identified sets of central neurons. So the question of how gene or protein expression influences neurophysiology and behavior is not such a pie-in-the-sky endeavor anymore.

You published a Nature paper back in 2015 about Drosophila crowd dynamics. Do you still work on that?

What fascinated me most from that study was the degree to which we observe differences across individuals. There were always one or two individuals that would initiate a cascade of group responses to the sensory cue. That observation really captivated my attention. From where do those individual differences arise? By understanding individuality across animals, can we better learn how to sculpt behavior for synthetic neurobiology? So my focus has really moved to the level of an individual’s nervous system.

You mention on your lab website that neurogenetics can have implications for artificial intelligence and for treating certain neurological disorders. Can you expand on that?

Neurological disorders are often very complex diseases. I think that studying differences in neuronal gene expression across individuals will be critical for understanding why certain people are more inclined to respond negatively to environmental stresses. One of the many advantages of working with Drosophila in this context is to be able to pin down mechanisms that define such individual differences at multiple scales ranging from circuit physiology to behavior.

As for Artificial Intelligence, when I worked in robotics as a postdoc I became very interested in determining how algorithms used by living systems can be leveraged to better control artificial systems. Drosophila has a very small and compressed nervous system that can nevertheless control many interesting behaviors using multiple limbs and wings. So if we can understand how roughly 100,000 neurons organize learning and memory, mating, fighting, courtship, walking, flying, grooming, all of these things in one small package, I think it's going to be transformative for robotic control.

What are your immediate research plans with EPFL?

I think understanding the control systems in the fly can have the largest impact at the moment. If we can understand how the nervous system coordinates complex behaviors, I think we can already start to think more clearly about algorithms for controlling robots and how neural circuits can differ across individuals.

So as a part of IBI I am very excited to join roboticists, computer scientists, and biologists to investigate neural networks in animals. We plan on imaging the activity of neurons that control complex limbed behaviors. We have some new tools that allow us to image the activity of these neurons, so we can really visualize their collective activity while animals make and act on decisions. In the longer term, we’d like to then figure out how the activity of these same neurons can differ across individuals or behaviorally divergent species.