4 ms·

Well, if you're looking for a textbook: Kandel is the gold standard, but it is quite hefty and technical -- a great slimmer one is "From Neuron to Brain" (adva

by psyklic 18y ago

Well, if you're looking for a textbook: Kandel is the gold standard, but it is quite hefty and technical -- a great slimmer one is "From Neuron to Brain" (advanced undergrad/early grad): http://www.amazon.com/Neuron-Brain-Cellular-Molecular-Approach/dp/0878934391 http://www.amazon.com/Neuron-Brain-Cellular-Molecular-Approa... If you're more interested in the computational end, try "Theoretical Neuroscience": http://www.amazon.com/Theoretical-Neuroscience-Computational-Mathematical-Modeling/dp/0262041995 http://www.amazon.com/Theoretical-Neuroscience-Computational... And I'll also add in a great computational textbook on spike recording: http://www.amazon.com/Spikes-Exploring-Neural-Computational-Neuroscience/dp/0262681080 http://www.amazon.com/Spikes-Exploring-Neural-Computational-... I'll also note that if you apply to a "Neuroscience" grad program, you're probably going to be doing biology experiments all day, or psychology experiments if you pursue cognitive neuroscience. If you want to use CS and Math, the best you can do in a neuroscience program is data analysis. Hence, I would recommend AI or Mathematical Psychology (though the latter is about modeling high-level events).

robg 18y ago
I agree on the first three and disagree on the last two points. I know neuroscientists that do few biology experiments, if at all. There are other ways to get data if you don't collect it yourself. I also know folks in Psychology programs that don't do psychology experiments, based on the same principle. For instance, different types of modeling allow for contributions that coincide with data but as theoretical and analytic tools. I wouldn't recommend Mathematical Psychology (too much theory, not enough application). I don't know enough about AI programs to comment.
- psyklic 18y ago
  I should have clarified further ... by and large, if you apply to a program in "Neuroscience," you are applying to a program in "Neurobiology." For example, MIT has a huge Brain and Cognitive Sciences option, with a beautiful new building dedicated to Neuroscience. However, a few years ago I was told they only accepted one new student who applied only for computational work that year, since they have very few professors who would support computational work alone. The fact is that there are very few professors working on "Computational Neuroscience," which is basically modeling experimental results. (I consider this data analysis because most of your time is spent analyzing experimental data. But, its intention is to not only explain given data but to predict the results of future experiments.) Hence, if you apply to a neuroscience graduate program, there will usually only be one or two faculty members who do computational work. (There are some exceptions -- e.g. UCSD, Caltech.) Many programs won't even take you if you're only interested in computational work because if you somehow don't get along with the professor or it doesn't work out, you won't have a backup lab. AI programs are divorced from experiments, but they get at the "bigger issues" and are satisfying for CS people -- whereas computational neuroscience models very complex small systems. Unfortunately, computational neuroscience also seems to be struggling -- many biologists discredit it, saying that its only real breakthrough has been Hodgkin and Huxley's model of the action potential!
  - robg 18y ago
    I'm at MIT. I don't think that's true. In my lab alone (neuroimaging), there are two students focusing on computational issues. True it's of data, but that's the challenge to any scientist - explaining the data in front of you and coming up with ways to collect better data. Modeling is just one form of that. Computationalists, therefore, have much to add.
  - yummyfajitas 18y ago
    You might want to look slightly outside the Neuroscience department. Math and CS departments also do quite a bit of computational neuroscience. Here at NYU, we've got at least 5 computational neuroscience people within the courant institute (math+CS depts), and there are several more in the actual neuroscience department. Plus we have a computational biology Ph.D. program within cims. I know Rutgers has the BioMAPS computational biology program, and I think princeton has something similar.
  - psyklic 18y ago
    I suppose that our main disagreements revolve about the ambiguous definition of "neuroscience," which is nowadays an all-encompassing term which has seemed to broaden to mean "absolutely any work which even mildly studies any aspect of the nervous system or behavior." In this respect, there are many opportunities for computation people. In neurobiology (the original "neuroscience"), the opportunities may be more limited. I was moreso discussing this latter definition in my posts, mainly because this is what many neurobiology-centric academic departments are called.

by psyklic 18y ago

robg 18y ago

I agree on the first three and disagree on the last two points. I know neuroscientists that do few biology experiments, if at all. There are other ways to get data if you don't collect it yourself. I also know folks in Psychology programs that don't do psychology experiments, based on the same principle. For instance, different types of modeling allow for contributions that coincide with data but as theoretical and analytic tools. I wouldn't recommend Mathematical Psychology (too much theory, not enough application). I don't know enough about AI programs to comment.

psyklic 18y ago
I should have clarified further ... by and large, if you apply to a program in "Neuroscience," you are applying to a program in "Neurobiology." For example, MIT has a huge Brain and Cognitive Sciences option, with a beautiful new building dedicated to Neuroscience. However, a few years ago I was told they only accepted one new student who applied only for computational work that year, since they have very few professors who would support computational work alone. The fact is that there are very few professors working on "Computational Neuroscience," which is basically modeling experimental results. (I consider this data analysis because most of your time is spent analyzing experimental data. But, its intention is to not only explain given data but to predict the results of future experiments.) Hence, if you apply to a neuroscience graduate program, there will usually only be one or two faculty members who do computational work. (There are some exceptions -- e.g. UCSD, Caltech.) Many programs won't even take you if you're only interested in computational work because if you somehow don't get along with the professor or it doesn't work out, you won't have a backup lab. AI programs are divorced from experiments, but they get at the "bigger issues" and are satisfying for CS people -- whereas computational neuroscience models very complex small systems. Unfortunately, computational neuroscience also seems to be struggling -- many biologists discredit it, saying that its only real breakthrough has been Hodgkin and Huxley's model of the action potential!
- robg 18y ago
  I'm at MIT. I don't think that's true. In my lab alone (neuroimaging), there are two students focusing on computational issues. True it's of data, but that's the challenge to any scientist - explaining the data in front of you and coming up with ways to collect better data. Modeling is just one form of that. Computationalists, therefore, have much to add.
- yummyfajitas 18y ago
  You might want to look slightly outside the Neuroscience department. Math and CS departments also do quite a bit of computational neuroscience. Here at NYU, we've got at least 5 computational neuroscience people within the courant institute (math+CS depts), and there are several more in the actual neuroscience department. Plus we have a computational biology Ph.D. program within cims. I know Rutgers has the BioMAPS computational biology program, and I think princeton has something similar.
- psyklic 18y ago
  I suppose that our main disagreements revolve about the ambiguous definition of "neuroscience," which is nowadays an all-encompassing term which has seemed to broaden to mean "absolutely any work which even mildly studies any aspect of the nervous system or behavior." In this respect, there are many opportunities for computation people. In neurobiology (the original "neuroscience"), the opportunities may be more limited. I was moreso discussing this latter definition in my posts, mainly because this is what many neurobiology-centric academic departments are called.