As a systems biologist who works on neuroscience projects and has worked in computational neuroscience I very much enjoyed the review. I found it accurately surveyed both fields and the disconnect between them. But I was especially intrigued with the final “Looking Into the Future” section and specifically with a few provocative but terse comments on which I hope Erik De Schutter may comment further.

In the Future section it is remarked that the two fields have different interests, e.g. that many computational neuroscientists have little interest in genes and molecules. I think this distinction is fundamental. Within the SBML community for example, there is some discussion of the long-term goal of scaling to the cell and beyond, but the center of gravity and pretty much all of the work is molecular. Conversely, while it is true as noted that kinetikit provides the opportunity for neuromodels to reach to molecular phenomena, the center of gravity is the electrical membrane behavior of cells organized in neuronal circuits.

There are a couple of consequences of this difference in focus. Electrical behavior is studied in the time domain, and thus its modeling naturally is as well. Time-series global molecular data is scarce. Systems biology modeling currently takes two approaches: (1) mechanistic modeling of signaling/gene regulatory pathways that involve a relatively small number of components but captures our current understanding of their detailed mechanisms and (2) data-driven modeling where data-driven refers to using high-throughput molecular data to construct empirical models because we don't have enough information to propose mechanistic models. In contrast, neural electrical behavior models, including the Hodgkin Huxley ones, are not mechanistic like (1) above but are empirical-phenomenological fits to electrical behavior, and are not like (2) above in not resting on global molecular datasets. So there is not a lot of common ground shared by the fields, although I accept the point that there could be going forward.

The review notes “a big disparity in funding levels”, that computational neuroscientists who can fit may be crossing over to systems biology, that computational neuroscience may gradually disappear, and finally questions the stability of the neuroscience field itself. I have also worried about these questions but given the gigantic powerhouse neuroscience appears to be they are contrarian. Nonetheless, for all of computational neuroscience’s successes and tradition, the field of neuroscience has not yet provided a stable review and funding mechanism for it. I have seen several of my senior computational colleagues who once attended neuroscience meetings now identify themselves as cell biologists. And the field of neuroscience appears an outlier in not adopting omic and systems biology approaches, rather maintaining a narrow focus on the search for particular disease genes. It worries me to see well-respected neuroscience friends proposing classical methods to “associate” the presence of an mRNA or enzyme with a neural disease when it is clear that any such particular molecule is going to be one of many hundreds so associated. Clearly this puts the field at risk of being upstaged by investigators outside the field. The review is correct that, at the limit in cognitive processes, neuroscience is distinct, but the vast bulk of the neuronal molecular processes under study now appear common to all mammalian cells and open to study by common means.