MAKING COMPUTER CIRCUITS WITH NEURONS: AN INTERVIEW WITH DR. RODNEY DOUGLAS

By: Joseph YaoContributing WriterThe human mind is often compared to the computer and sometimes a computer is compared to the human mind. While the two can perform similar functions, the human brain is more complex, which is apparent in the way the mind is formed. While computers are made by an external source, the human mind is unique in its ability to form itself from internal factors. The “self-construction” of the human mind has many layers of complexity that are explored in the sciences of Developmental Biology, Cognitive Science, Neuroscience and Physics. Dr. Rodney Douglas recently spoke at the Rockwood Memorial Lecture on his work in unraveling how to model the way the brain makes computer-like circuits with neurons.Douglas works for the Institute of Neuroinformatics at the University and ETH (Swiss Federal Institute of Technology) of Zurich, and conducts research projects regarding issues at the cross between neuroscience and information science. He has published numerous papers in scientific journals such as Science and Neural Computation, and has showcased his research at multiple international science conventions.In the lecture, Douglas introduced his research on Constructive Cortical Computation. Douglas has been creating computational models that can instruct virtual cells to reproduce in the same way that scientists have observed brain cells reproducing during the development of the human brain. His research attempts to establish the relevant genetic markers and cellular instruction manual for the developing neurons that will eventually lead to the formation of the multiple layers of cells in the brain’s cortex. In addition, Douglas is interested in the wiring of circuits made of neurons that is established at this point of development, specifically the complex branching behavior that is common in certain types of neurons.While Dr. Douglas’ personal research does not pertain to biological or health care related issues, his groundbreaking research can lead the way to further developments in biomedical technologies. If full construction of cortical circuits can be achieved, this research can highlight key effects of neurological disorders, especially disorders of development. By transferring the genome into a computational code, point mutations can be intentionally made to specific genes and the effects of those mutations can be simulated and analyzed. Not only will this lend insight to the cellular details of neurological disease as the brain develops, the effects in other biological systems can also be “decoded”, and DNA can possibly become computer software: a program to simulate and analyze diseases. And in that far future, if DNA can be encoded into software, then the reverse could possibly be true: where software can be transformed into DNA, and DNA into life. Douglas’ work has scratched the surface to a world where science and life are connected in ways not yet imagined.After his lecture, PROSPECT was able to speak with Dr. Douglas on his interests and research.PROSPECT: How did you become involved in how the brain sorts information, and what about this field interests you the most?DOUGLAS: I have always been fascinated by the question of how the brain is able to process information. It has been my fascination since I was 18, so it is difficult to decide exactly what happened. But surely, I realized that in order to do that I would have to understand cortical circuitry. It seemed like obvious materialistic position, and at a certain point I was able to do that.PROSPECT: And how did your interests expand from cortical circuitry to cortical construction?DOUGLAS: So I have been working with Kevin Martin, particularly on the question of identifying and characterizing cortical circuitry. Some years ago, we published for the first time a connection matrix for a cortical area. And in thinking about what the implications of that circuit are, it becomes clear that the way it operates must be deeply connected to the way it constructs itself. So for example, we consider, or very often people consider, learning and cortical circuits to be learning on a completed circuit; so you imagine a circuit is given to you and then you deal with the problem of learning. Where it seems to me that, more and more, learning is something that is part of a process of setting a configuration that plays out over construction smoothly into something that we call learning in relation to the outside world.PROSPECT: That is very interesting. I could not help but think, during the lecture, about how one day you can take DNA and form it into a code.DOUGLAS: That is the vision, but at the moment there is a lot of detail in the genetic level of experiments. But biology does things sometimes in a complicated way for reasons that are not entirely clear, and you find that you can reduce the principles. When you look across species, it is very easy to see some commonality of function even though the shapes and stuff are very different. So the hope here, or the expectation, is that one can take a higher-level engineering view to achieve maybe not the same complexity as biology but the same principles, in just the way that you pointed out.PROSPECT: With that in mind, what types of obstacles do you see for your future research?DOUGLAS: Death? [laughs] Well, so you saw the work that I presented in the lecture has no material foundation. Of course there is experimental foundation, but the ideas are necessarily played out in a simulation environment. And that is somehow unsatisfactory. It is good because it enables you to get a grip on the principles and to examine the implications of those principles, but it is not good in so far as there is not an experimental system in which you can evaluate this. Although my interest is somehow in circuitry that produces comparable, or can emulate, biological computation, it is not a medium where you can do those types of self-organization. So one will have to look somewhere else, and the obvious bet is that, more and more, we see how cellular biology can be manipulated to one’s ends. And you look at projects like the Amgen project at MIT, where students are already somehow “piggybacking” on the genetic machinery of cells. I think that technology will move towards cellular technology, where you strip down cells with limited functionality to produce cell lineages of your choice. And then in that far distant future we can see where a theory and a biological technology can come together. So my feeling is it is in that direction. And if you look at the kind of work done by Greg Venter, last year in his publication, on producing or replicating cells with artificial genome, there you begin to see that type of future very clearly.PROSPECT: What form of impact do you see your research will have on healthcare?DOUGLAS: Well, this is a tender point for me because I am sure all of Neuroscience is useful and should be used for health purpose, but my particular interests is in developing technology. So of course if you understand how cortical circuitry proceeds, then you have a principle (if you understand it from a computational point of view and not a developmental point of view) with insights on how you can modify developmental process. And this surely has an application in health. However, my research is cast at a level where the concern is for “engineering” understanding rather than for “biological” detail.Photo Courtesy of Hljod Huskona