Cross-Pollination at Cosyne 2018

The 15th annual Computational and Systems Neuroscience conference (Cosyne) was held for the first time in Denver, Colorado, from March 1-4, followed by two days of workshops in the mountain resort town of Breckenridge. While the location was new, the theme of bringing computational and experimental neuroscience together remained as strong as ever.

This was evident not only in the parade of talks and posters at the main conference, but also more explicitly in some of the additional events, such as the Simons Foundation Computational + Experimental Matchmaking and the workshop entitled “Manifold-Splaining: What the Theorist Said to the Experimentalist.” This kind of interdisciplinary communication has always been a hallmark of Cosyne. As Linda Wilbrecht, one of the program chairs, said during the closing remarks of the main conference, “People who show up [at Cosyne] have the generosity to reach across boundaries.”

With 800 attendees, Cosyne is a small conference compared to the behemoth Society for Neuroscience meeting, which draws tens of thousands. But Cosyne’s single-track design allows it to cross boundaries more effectively than broader events like SFN. Given the wide range of topics represented at Cosyne — from perception to probabilistic models, from dendrites to deep learning — the presentations fall outside the scope of most attendees’ expertise. Forcing everyone into the same room brings together not just theorists and experimentalists but researchers studying different species, brain areas, tasks and techniques.

This effect of small conferences is superbly highlighted by the neurogeneticist and Nobel Laureate Martin Chalfie’s description of the early days of C. elegans meetings, first launched in 1971:

The entire field was there, all 125 of us, and people worked on a wide variety of problems; after all we had an entire organism to study. This meeting set the precedent for the C. elegans meetings for the next 10-15 years: we all attended and were interested in every session. Because we wanted to understand this animal, talks on muscle function, cell death, chemosensation, cuticle formation, meiosis, cell migration, and many other topics were all equally interesting, and were often useful in our own work in surprising ways. … Since many of these areas were new to us, we received a wonderful general education, and we taught each other. In fact, those of us who attended that first meeting resisted as long as we could the idea of having multiple concurrent sessions at our meetings as the field continued to grow. Now, however, people go to sessions on the nervous system or embryonic development or cell biology and miss out other fields. Many of us, however, regret that we cannot learn about all the advances, especially those in areas far from our specific studies.

For an upcoming episode of the “Unsupervised Thinking” podcast, my co-hosts and I decided to explore this theme of scientific cross-pollination at conferences. We asked Cosyne attendees to tell us about any result or research method outside their own immediate area that they found particularly exciting or interesting. We collected 11 interviews from attendees from different backgrounds, ranging from doctoral students to professors (to hear the full audio of these, check out our next episode, which will be online at the end of March).

The responses we got were broad, reflecting the diversity of the people we sampled and multiple interpretations of the question. But certain themes emerged. Many attendees were excited by the Gatsby lecture, a special spot by an invited speaker on the first night, given this year by Iain Couzin. Couzin’s talk, “Collective Sensing and Decision-Making in Animal Groups: From Fish Schools to Primate Societies,” explored different instances of the complex group behavior that can arise from simple interactions among individuals. This talk has two interesting implications for neuroscientists. First, it highlights the importance of collective naturalistic behavior for understanding the behavior of individuals, which tends to be a greater focus of neuroscience. Moreover, analyzing collective behavior among individual animals is strikingly analogous to how we analyze the collective behavior of neurons based on their individual interactions. Seeing the tools of neural analysis applied at this higher level offered attendees new ways to think about these tools and about neural activity itself.

In a similar vein, an invited talk by Carina Curto entitled “Emergent Dynamics From Network Connectivity: A Minimal Model” generated buzz for its clean demonstration of relevant results from applied math. Curto described how interesting dynamics could emerge from networks that are simple enough to remain mathematically tractable. This talk was more technical and less neuroscience-specific than most at Cosyne, but it was all the more well-received for that. Presentations like these — which push the limits of topics normally covered by a conference — can be a refreshing reminder of alternative approaches to old problems. And this year, even more than in the previous years I’ve attended, Cosyne provided a good blend of solid science and surprise.

The excitement that these out-of-field talks generated among Cosyne attendees should send a clear message to conference organizers: Exposing your attendees to science just outside their comfort zone can invigorate them in unexpected ways. Such talks aren’t mere novelties; if chosen well, they generate enthusiasm and inspiration.

Another idea that emerged from our interviews was the parallel between communicating with scientists in other fields and communicating with the general public. Some of the findings that scientists cited as interesting outside their own area were directly related to human behavior, such as the neural representation of the sex of others, and how sexual experience alters that representation as well as the animal’s behavior. These topics also tend to attract the attention of lay audiences and can serve as motivation for people to enter the field of neuroscience in the first place. In this way, diverse conference interactions can help researchers zoom out and rediscover their own motivation. Personally, I know the broad reading I’ve done for our podcast has kept me excited about being part of the neuroscience community.

Of course, listening and speaking beyond our own fields presents challenges. Scientists are used to being very well-informed about their area of specialization and may not feel comfortable engaging with researchers in different fields. This is where good communication skills between scientists become essential. I know it can be tempting in a 10-minute talk to cut out some background slides to squeeze in extra results for the aficionados in the audience. It’s up to a conference and to the individual speakers to decide whether to preach to the choir or try to reach a broader audience.

Attendees also face time and energy constraints, even at a single-track conference. Taking in everything that a conference like Cosyne has to offer can be overwhelming, and most veterans have learned to acknowledge their limits and be selective about what they choose to see. In the end, a balance must always be struck between breadth and depth of information for a scientist to be able to make progress. As Santiago Ramón y Cajal said in his “Advice for a Young Investigator”:

Like unmolded steel, our mind represents a potential sword. The forging and polishing of study transform it into the tempered and keen scalpel of science. Let us have a cutting edge on only one side, or on two at the most, if we want to conserve its analytical powers and penetrate to the heart of problems.

Grace Lindsay recently completed her PhD at the Center for Theoretical Neuroscience at Columbia University. For the past two years, she has been co-host of Unsupervised Thinking, a podcast on neuroscience and artificial intelligence.