Sharing space to support ‘better science’

Across disciplines, researchers in the Computational Neuroscience Initiative put their heads together to better understand the brain.

Computational Neuroscience Initiativ
Vijay Singh (center), a postdoctoral fellow involved in the CNI, discusses his research with others involved in the initiative.

In 2010, Penn professors Vijay Balasubramanian, Joshua Gold, and David Brainard had an idea. They recognized that the University had a rich landscape of researchers in disparate schools and departments who were investigating related questions, all dealing with computational neuroscience and how the brain works. They wanted to create an initiative that would build upon that strength.

“There were many of us on campus who were doing somewhat related computational things, and we would see each other occasionally, but we wanted to see each other a lot more,” Gold says. “There’s just this synergy that happens when you share a space with people and are able to talk to them every day rather than every couple of months. That’s how ideas get shared. It just seemed pretty clear that if we really wanted to make computational neuroscience move forward, it wasn’t a matter of bringing in new people but rather consolidating the people who were already here.”

Computational Neuroscience Initiative
Vijay Balasubramanian (left), Josh Gold, and David Brainard (not pictured) started the CNI with the goal of creating a shared space where faculty, students, and postdocs could share ideas related to computational neuroscience and the brain.

Balasubramanian, Gold, and Brainard wanted to create a space that would encourage discussion and collaboration between these different researchers, which is how the Computational Neuroscience Initiative (CNI) was born. With help from the School of Arts and Sciences and the Perelman School of Medicine, the professors set up a shared space in the Richards and Goddard laboratories on campus where faculty, students, and postdocs could share ideas.

“I don’t know whether it’s Penn’s history or its small campus,” Brainard says, “but there’s a lot of interaction between faculty here. Interactions across departments and schools are very porous at the faculty level. There was already a lot of collaboration happening between various pieces of the group, but creating the CNI made it easier to facilitate this effort.”

The questions tackled by members of the CNI have implications for everything from neurodegenerative diseases to machine learning and artificial intelligence.

“Neuroscience has been around for a hundred years,” Balasubramanian says, “but in some ways the field is still in its infancy. The brain isn’t just made of one or two neurons—there are a hundred billion interacting neurons. So you need to understand how they work collectively to produce emergent phenomena such as perception, cognition, and sensation. It’s only now, in this epoch, that amazing tools are coming to light for interrogating and understanding the behavior of many, many neurons at once. Building an understanding of the collective behaviors of many neurons acting together is going to absolutely revolutionize what we can do both technologically and medically.”

The CNI includes professors from many departments in Arts and Sciences, Medicine, the School of Engineering and Applied Science, and other schools at Penn. According to Gold, the field of computational neuroscience is about using computational and theoretical tools to better understand the brain.

ribbon synapse
A ribbon synapse from a rod photoreceptor to its neighboring cells. David Goodsell, from Philip Nelson, From Photon to Neuron: Light, Imaging, Vision (Princeton, 2017).

“It can really be on three basic levels,” Gold says. “One level is using mathematical and statistical tools to analyze data from the brain. Another is building computational models of the brain. The third is building devices to interface with the brain. So it spans lots of dimensions. I think a huge advantage of the CNI is that all of the members are both theorists and experimentalists. It’s not like we’re building mathematical models that exist only as mathematical models, we’re all interested in applying the models directly to data.”

According to Gold, one of the defining features of the CNI is having postdocs who are not hired by a specific lab but by the center itself to provide a lot of the “collaborative juice” and develop computational expertise. They also build bridges between various lines of research. The CNI currently has four postdoctoral fellows: Gaia Tavoni, Vijay Singh, Eugenio Piasini, and Eve Armstrong.

It’s not like we’re building mathematical models that exist only as mathematical models, we’re all interested in applying the models directly to data. Penn Professor Joshua Gold

“The CNI postdocs are the centerpieces of our whole program,” Balasubramanian says. “In modern neuroscience and the science of the brain, there many things that require the combined expertise of different groups, especially theorists and experimentalists. As biology becomes more and more quantitative, a whole group of people with specialized analytical skills is emerging. Our postdoctoral fellows are chosen because they have these skills, and are naturally placed to work jointly between multiple labs and groups, providing the glue to hold it all together.”

Singh, who has a physics background but is interested in the brain, neuroscience, and biology, says he wanted to work in a place where he could use his tools from physics but at the same time talk with people who knew more about biology so that he could collaborate with them and apply his knowledge of physics to help understand new problems.

“One thing I enjoy about the CNI,” Tavoni says, “is that people are very friendly and passionate about their research and are always willing to discuss topics related to their own or other people’s research. It’s a very open environment where we can have interesting discussions on a daily basis. I also really like that we are given a lot of freedom to explore the questions we are most interested in.”

Computational Neuroscience Initiative
This image, a variant of an optical illusion called Adelson’s checker-shadow illusion, was created by Brainard’s group, which studies human vision, both experimentally and through computational modeling of visual processing.

The topics that the CNI tackles range from decision-making to visual, auditory, and olfactory sensation, and these processes work together to enable affect our cognition and perception of the world. The potential applications touch topics in medicine, allowing for things like the earlier diagnosis of neurological disorders and ocular diseases as well as the development of tools to better treat them.

“We are what we think,” Balasubramanian says, “and there are many ways in which people develop disorders of the mind and brain that change the way they think. The first step to knowing how to treat those kinds of disorders is to know how the brain is supposed to work to support the normal functions. You need to understand how the circuits should work in order to fix them.”

The research also has the potential to revolutionize technology. The knowledge gained about the networks in the brain, neurons, and their connections can lead to developments in machine learning and artificial intelligence.

There are so many people at Penn who do complementary things, Gold says, that just having a structure in place that allows them to build on those relationships makes the whole more than the sum of its parts.

“There’s a lot of attention being paid to the rigor and reproducibility of science, a lot of which has to do with quantitative techniques,” Gold says. “Having a place where our expertise is consolidated, instead of having that kind of helter-skelter across campus, allows us to build an infrastructure that helps support better science.”

Homepage photo: One of the defining features of the CNI is having postdocs who are not hired by a specific lab but by the center who provide computational expertise and build bridges between various lines of research.