On Monday, the New York City-based group NeuWrite hosted a public debate on minds, maps, and the future of neuroscience between Sebastian Seung of M.I.T. and Anthony Movshon of NYU, moderated by Robert Krulwich (Radiolab) and the esteemed science writer Carl Zimmer (NYT, Discover). As eager attendees packed Columbia University’s Havemayer Hall on Monday evening and another three hundred watched a simulcast from a nearby room, two things were immediately clear: there is a hunger for a true debate about the brain, one that moves the conversations usually held behind closed doors at scientific conferences and over late-night beers to the public sphere, and Sebastian Seung is wearing gold sneakers.
Credit for organizing the event goes to NeuWrite, an innovative and resourceful group of scientists, writers, and, as their website explains, “those in between: graduate, post-doctoral and faculty researchers, fiction and non-fiction writers, as well Journalism and MFA students at Columbia.” NeuWrite regularly workshops pieces of print journalism and books-in-development with a scientific focus, as well as film, radio, and poetry that present threads of scientific inquiry. In public, this is NeuWrite’s second event about the brain—last year paired Patricia Churchland, author of Braintrust, with Roger Bingham of UCSD and Jesse Prinz of CUNY, for an engaging discussion about morality and neuroscience.
It was clear from the opening statements at Monday’s debate that Movshon and Seung represent two different schools of thought, but their conversation ended up being less a “brain brawl” and more a respectful airing of differences. Seung believes neuroscience is stuck in a traditional mode of research, where the necessity to publish the next paper and get the next grant corrals scientists into overly-specific, limited fields of view of the whole system they’re studying. As a result, Seung argued, “neuroscientists can be very short-sighted.” Seung’s own plan of attack is one he’s elaborated in his popular TED talk and documented thoroughly (and very accessibly) in his new book, Connectome: How the Brain’s Wiring Makes Us Who We Are. On Monday, he reiterated this philosophy: the best way to understand perception, memory, and the basis of psychiatric disorders like schizophrenia and autism, Seung believes, is to study the brain at the level of the synapse—to trace all the connections between all the neurons in a brain. By generating a map of the whole system, we may be able to finally see engrams for memories and perceptions, as well as what might be going wrong with these networks in the aforementioned disorders, perhaps due to various problems in the ways neurons are wired up, which Seung calls “connectopathies.”
So where’s the debate? Movshon made his position clear: “I’m not going argue against the acquisition of information. I just don’t think the connectome is the way to do it.”
Movshon then presented some of his concerns about Seung’s connectomics. Among them:
- There is a scale mismatch between the microscale field of view that tracing a connectome gives you (you’re looking at connections between cells of one specific organism) and the mesoscale understanding that Movshon argues is what’s really needed to understand the big questions (the mesoscale being statistical probabilities of wiring and activity common to the different brains of individuals).
- The relationship between the computations carried out by a brain and the substrates of those computations remains elusive. In other words, even if we can see the connections between all the neurons in a brain (the substrate), how can we be sure that we’ll then make the leap to understanding how those connections give rise to a perception or memory (the computation)? Movshon brought up work done by Todd Sacktor that suggests there may be molecular switches within neurons and at the synapse that play a major role in the maintenance of memories. Movshon argued that connectomics would not show us the potentially crucial molecular mechanisms such as those studied by Sacktor (for more on that work, here’s an interview I did with Sacktor on a past edition of our podcast).
With their differences stated, some of the most intriguing moments of the evening arrived nearer to the end of the debate. Krulwich turned to Movshon and posed an important question: if you don’t want to map the connectome, how are you going to understand the brain?
Movshon responded by reminding the audience that “neuroscience is a cottage industry,” meaning the study of the brain has traditionally been carried out by many individual labs focusing on different parts of the whole, communicating their results to the scientific journals for circulation to the other cottages. In this philosophy, which Movshon believes is still the best way forward for the field, a better understanding of the whole brain and the answers to the big questions of perception, memory, and disorder will emerge from a better and better understanding of all the parts– as carried out by the localized cottage industries– and eventually consensus will emerge about the whole.
Seung’s approach is one injected with a bit more grandeur—and Movshon pointed out that “The problem of grandeur in neuroscience is one we’re all concerned about.” Connectomics is a large-scale undertaking that, a bit like large-scale brain simulation projects, demands rapid, parallel improvements in computer technology. Connectomics does not follow the research traditions of the cottage industry that Movshon represented in the debate—though the questions connectomics could answer are indeed big ones, no one can be totally sure yet exactly how those questions will be answered, or when. It’s a very different research approach than, say, studying fear response in the mouse brain for one’s entire lifetime.
The discussion of grandeur in neuroscience inevitably led to a briefly contentious back-and-forth about Henry Markram’s Blue Brain Project. When Movshon brought up Blue Brain and seemed to suggest a parallel between Seung and Markram, arguing that what we really need are more “guided, more focused, and more hypothesis-drive projects,” Seung pounced on the chance to distinguish the aims of his work from those of Blue Brain. While Markram famously declared in 2009 that he would have a full human brain simulation completed within ten years, Seung takes a somewhat humbler approach to his work: “Hey, I just want to map some connections,” he joked, draining the tension out of the hall with a good laugh. Indeed, Seung is hoping that in his lifetime he can map a mere cubic millimeter of mouse brain—just that, he said, would be a big step forward for the field.
Though there was no blood spilled in the end, one can hope that this “brain brawl” will be the start of more such public discussions of neuroscience’s direction and goals in the 21st century. The difference of approach between Movshon’s cottage industry and Seung’s connectomics is not necessarily one of ambition but one of scale. Both scientists emerged as ambitious explorers using slightly different tools and drawing slightly different maps.
These questions of approach and scale may present more two-camp issues as fodder for future debates (it could also be interesting to see such a debate unfold in more of an Oxford-style, Intelligence Squared format). While the actual science happens in labs and is reported on in journals, at conferences, and eventually in the press, we rarely get to hear actual scientists talk to the public and to each other about why it is they are going about studying the brain in the way they’re studying it. One can only hope Monday evening was the first of many such synapses between the scientific community and the eager public.