Partha Mitra is the Crick-Clay professor of biomathematics at Cold Spring Harbor Laboratory in Cold Spring Harbor, N.Y. A theoretical physicist by training, he is working on mapping mouse brain circuits. We spoke on the phone Tuesday evening about President Obama's $100 million initiative to map the human brain, which he had announced earlier that day at the White House. A lightly edited transcript follows.
Dylan Matthews: You were critical of the [Obama Administration's] initial brain-mapping proposal. What did you make of the details that came out Tuesday?
Partha Mitra: It was a significant improvement compared to the early media reports, where some questions arose about what exactly was being proposed, where the money would come from and so forth. I think what we heard today was a fairly moderate proposal. And it wasn't so much a proposal; rather it was a proposal to make a proposal. It was $100 million in fiscal year 2014 to develop appropriate technologies, and then there is a promise of more funding for a more ambitious project, still to be defined through a deliberative process. That’s different from what was initially talked about, and better.
Matthews: Is that how you'd spend the $100 million, if you were taking charge here?
Mitra: $100 million seems like a lot of money, but the sequestration has taken billions out of the fiscal year 2013 science budget, so there's been a huge hit across the board on research funding. So there's a question as to whether that sequestered science funding is going to be restored or not. One really hopes this will be the case. Otherwise there is going to be a big negative impact on all science research including neuroscience.
Assuming that funding is restored, that's a different picture. If it is not restored, I'd imagine this money should at the very least support projects that are already going on that could advance the basic agenda. This is not the first time people are making ambitious initiatives for studying the brain. Those proposals that have been made in the past also require support. So while there's a sense that, "Oh, we're going to propose something new," there's already ongoing research -- there's a lot of people working very hard and trying to get grants.
There have been specific discussions about newer technology. There were no specific science goals in the project as laid out except to say we need to study the brain because it is exciting and not understood. Those of us who work in neuroscience would most certainly agree with that sentiment.
There has been a push for better technologies to make recordings from many neurons. That's definitely necessary and sensible. Whether one should spend $100 million on that, I don't know, and it doesn’t sound like that is what is being planned now. Presumably one would spread that funding more evenly among more than one avenue of research, and I would imagine this would emerge from the meetings that are going to happen.
Matthews: One suggestion I heard when this was initially proposed was that we should start with smaller mammalian brains and then work our way up to humans.
Mitra: That's perfectly fine, and that's what people are going to do. Again, when this was first said, and no real details were given, people sort of filled in the details. And some of what they filled in was very unrealistic.
There was a statement in the New York Times about recording from every neuron in the human brain. That doesn't make any sense. Even if you took a whole mouse brain, that still isn’t workable. There is no real possibility of that when you look at the basic physics of how light scatters in brain tissue or the lengths of antennas you need to make radio transmitters.
What we have subsequently heard are moderate, more realistic goals. Going from recording hundreds of neurons to recording thousands, or tens of thousands, okay, that sounds doable. Under what circumstances and how it will be done, all this will really have to be discussed and developed.
It is not the first time people are recording from many neurons. There was a lot of excitement about this in the 1990s that people may not quite remember. Like many neuroscientists who were participating in that excitement, I've spent time analyzing many neuron recordings. I was fortunate enough to have co-directed a summer course on analyzing multichannel neural data for many years at the Marine Biological Laboratories. It's an ongoing research topic.
Scientists study whatever organism is easiest technically. For example, people have recorded simultaneously from many neurons during the swimming behavior of the leech, or from the gut ganglion of the lobster and other invertebrates. Recordings have been made routinely from hundreds of electrodes both in rodent and primate brains. There is some excellent recent work monitoring neural activity across the whole brain in zebrafish larvae. As technology improves, we can explore more species in a productive way. A big advance was the development of better optical sensors and optogenetics and, more generally, tools from molecular biology that came into neuroscience. We should keep developing technologies, and they'll improve over time.
DM: You're also skeptical of the comparison to the Human Genome Project.
PM: One thing I see lacking is a clear answer to, "What is the scientific goal one is pursuing?" There were analogies made with the Human Genome Project and mapping brain activity that I don’t understand. Sequencing the genome is a different kind of goal from saying, "Let's measure from many neurons." If one must make analogies (and this is a dangerous thing to do because of plasticity, development and individual variation), it's probably structural neuroscience, the physical arrangement of the circuitry, that's a little closer to what the genome is. Brain activity comes as a result of the structure of the circuit plus environmental inputs. If you look at something, there's neural activity in your visual cortex. That's a product not just of the cortex but also of what you're looking at.
There has been an evolutionary dynamic to the genome, but [barring cancers] it's a reasonably well-defined object. It doesn't vary that much from individual to individual in a species. So if you must draw an analogy, it would be the circuitry. And that doesn't just include the synaptic connections; it would also include the physiological characteristics of the neurons, cell types and so on. Those things are a little more stable. The activity pattern is contingent and varies based on inputs.
So the definition of the scientific goal is not quite there yet. It's more like, “Let's develop tools.” And I think that's great, I do believe we need better tools. One should have goals and develop tools to meet those goals. But someone like me, who has a theoretical background, wishes there'd be a bit more discussion of what our scientific goals are.
Matthews: A rather applied goal that gets mentioned is getting a better understanding of what's going on with neurological disorders like Alzheimer's or Parkinson's.
Mitra: Right, that's a rationale for doing it for a large extent.
Matthews: I don't know how directly actionable that is in terms of a basic science agenda, but what avenues of basic research do you find most promising in terms of that medical goal?
Mitra: That’s a big question, and clearly we don’t have the answer. On the one hand we have the real imperative to study brain disease, and on the other hand we're talking about better tools to record from many neurons. These are both very important, but one still has to connect these two dots. One thing seems clear, that many different tools are needed. One needs the tools being talked about, but in addition one may need tools that don’t have anything to do with activity measurements.
Take Alzheimer's. Now I’m definitely not a domain expert, but it seems that scientists don’t yet fully understand the etiology of Alzheimer's. So what is the mechanism that is giving rise to that? Let's ask that question. One approach that has in fact led to insights about the possible disease mechanisms in the past is looking at the microscopic anatomy and histology of postmortem Alzheimer’s brains. If we had better tools to study circuits in postmortem human brains, then perhaps we’d gain some important clues into Alzheimer’s disease. It’s definitely a technology in the need of development. We’ve seen the increased concerns about traumatic brain injuries in sports, another area where we lack a full understanding. By now some brains have been studied post-mortem to try to understand the etiology. There's an example where there's no activity involved, yet there are clues and hints as to what might have gone wrong. I’m not saying that this will replace activity measurements, just that brain activity measurement is only one part of the story that connects the dots between disease and basic scientific understanding of the brain.
Understanding what's going wrong in the brain in neurological disorders is a very important goal, because we don't really know in many cases. People are working very hard on these problems and that's why there needs to be more of a dialogue, asking, "What are the technical problem that needs to be resolved?"
Once that sort of consultative process begins and there is an engagement, I'd say that there is hope for fundamental impact. That’s the role of the dialogue. One can't really decree a scientific agenda. People have to engage in it and participate in it.
DM: What should we want out of a federal neuroscience agenda like this?
PM: If I were to put out one word that's really critical, it's integration. Brain research is fragmented, people tend to do their own things. An important reason is the complexity of the subject – one can spend an entire career investigating a particular detail. It may also be due to the competitive funding mechanisms, with people focusing on their own laboratories. With these real life constraints, how do you collaborate and integrate in practice? One way to do it is to understand, "What are the results that appeared in different fields for a particular overall scientific or medical goal?" I think integration is very important, therefore shared scientific goals are important. The tools should ideally follow rather than drive those goals.
Even though it might seem obvious that the goals come before the tools, it appears in practice the tools are coming before the goals. If we can’t agree on the goals, we end up agreeing on the tools. And there's where I think some people have questions. My question was simply, “What is the goal?” If the answer was “record from every neuron?”, why that particular goal? But I'm much more hopeful from what I heard today that there'll be a reasoned discourse, and meaningful goals will come about.