Neuroscience + Computer Science = ????

November 4, 2007 4:22 PM

What are my options for combining neuroscience and computer science in a graduate school setting?

I am about to complete my undergraduate degree in neuroscience and am in the process of applying for PhD programs in the same; however I am having serious second thoughts. I do not think that I want to do conventional lab work and am leaning more towards computational neuroscience, possibly in the neural networking/A.I. realm. Caveats: My computer science background consists only of courses in high school and a few dubious skills picked up along the way on my own. What type of program should I be looking for that will give me a good grounding in computer science (and accept someone with my background) but allow me to continue with neuroscience? Am I looking for computational neuroscience? Bioinformatics? Cognitive science? Masters or PhD? Should I be getting lab experience first as a research assistant? Do I need to do some kind of post-bac program in CS first? Help me plan my future!

posted by mayfly wake to science & nature (15 comments total)

As one who would love to be in your position, I can only add a "yes, how to proceed?".

Oh, and if I was to re-do my college career, I'd try really really hard to get into Carnegie Mellon. They're second only to MIT for computation/neuroscience/cognitive science.
posted by whycurious at 4:44 PM on November 4, 2007

I was in almost exactly your position one year ago. I chose to delay graduation to get a computer science minor. I'm happy with my decision.

You may not be able to do this. In any case, definitely get lab experience now. Three reasons: 1) You'll gain a valuable grad school reference. 2) It is easy to find programming/systems tasks as a research assistant. 3) You'll find out whether or not research is the career path for you.

I worked in a psych lab during my junior year of college, and then my professor asked me if I was interested in doing programming. He gave me a simple Perl task, and it went well. As time went on, I did more programming/system odd-jobs for the lab. I realized that programming, not research, was what I actually wanted to do. So now I'm planning to go to grad school for CS.

Grad school is a lot of work, and as opposed to college, you really have to mean it. By all means, take your time in deciding.
posted by Laugh_track at 5:02 PM on November 4, 2007

I would strongly advise you to ask these questions of your instructors during their office hours.

American PhDs are long. Avoid doing a master's degree before the PhD if you can. And you probably can.

Programs might be looking for people who are strong in either the biology or computational side of things, but having evidence of some proficiency in the other area would be a big plus. Having some experience doing computational neuroscience work as an undergraduate now would probably be the best thing that you could do. I presume you have a strong neuroscience/biology transcript—adding a glowing recommendation letter that says how quickly you picked up computational techniques would impress any admissions committee.

A general bioinformatics program probably won't be the best thing if you are specifically interested in computational neuroscience.

Good luck, from someone who also realized that biology was definitely what they wanted to do but wet lab work was not.
posted by grouse at 5:05 PM on November 4, 2007

Oh yeah, one more thing: if you want to make computer brains, take some time to reflect first and ask why.

Just because, ya know, I'm really hearing myself in your question here.
posted by Laugh_track at 5:12 PM on November 4, 2007

First of all, you've picked a good time to do computational biology-type work. The field is exploding, and there are a ton of interesting problems.

At most institutions, there's still a fairly wide gulf between the type of computational work you're talking about and the biology-driven wet labs. If you entered a CS department, you'd be developing algorithms, proving theorems, etc. It's all very math-y. From the biology side, you'd be munging data, and trying to model the systems you're looking at.

There are a few truly interdisciplinary labs, and since you're the neuroscientist, you're in a better position to identify those than I am. Get a hold of the latest neuro journals and scan them, looking for people who are doing work that interests you. Read through their papers, and see what kind of techniques they're using. Then, evaluate your own skillset to see where the deficiencies are. Then, immediately do everything you can to compensate for those deficiencies. What you'll have to do depends greatly on what kind of lab you're shooting to get into.

- If you choose something neuro-heavy, you may already be qualified enough to get into the school. This may also be true at grad schools where you don't pick a lab until your second year. This will give you time to take some extra classes, focusing on CS and theory, and then move into a lab where you can put those skills to use.

- You may have to take some CS courses now, and maybe even stay an extra year in undergrad to pick up a CS minor.

- You might want to do a masters program in either CS or Comp Neuro as a starting point. This, again, will give you time to get your skillset up to speed and make you a more attractive applicant when you apply to the labs that you're looking at for your PhD

In summary:
1) Figure out what you want to do
2) Figure out what skills you need to do it
3) Develop those skills
posted by chrisamiller at 5:21 PM on November 4, 2007

And yes, yes, yes - get research experience NOW! The number one thing grad schools like to see on your application is research experience. This shows that you know what you're getting into and that you're capable of doing it. It will also get you a fantastic letter of recommendation from someone that you've worked closely with.
posted by chrisamiller at 5:23 PM on November 4, 2007

Thirding the research experience, a friend I went to undergrad with (Computer Science degrees) went on to get his PhD in neuroscience, hated the research aspect, and is now a management consultant. And I wouldn't wish that on you.
posted by minedev at 5:36 PM on November 4, 2007

Talk to profs in neuroscience, and profs in CS, about this. They will know your situation better (even if you don't know them personally you can email them, say you'd like to make an appointment to talk, give them a capsule sense of your preparation so far and your goals, and let them know you want to talk about advice for grad work), and can talk you through the strengths of various programs at different schools.
posted by LobsterMitten at 7:58 PM on November 4, 2007

That is, talk to profs at your current institution about this, in person.
posted by LobsterMitten at 7:58 PM on November 4, 2007

A few things off the top of my head:

1) Neuronal simulation. Working with and writing programs that try to simulate how a neuron works. Check out NEURON. The user groups forum might potentially put you in touch with researchers who may be looking for grad students.

2) Modelling biophysics of spine motor control. Sorry, no links, but there are people working on how the physical body is controlled by the spine using merely feedback loops. Some decent money to be asked for (ie., simulating realistic movement for movies/games).

Sever a cat's brain from the spinal cord. Put cat on treadmill. Cat will walk.

Sever a frog's brain from it's spine. Tickle a spot on the back of the frog and the frog will 'scratch' that exact spot with a hind leg.

3) Bioinformatics - ie., data mining. I know some people who are working with the large databases and mining it for data. I'm hazy on this, but it has something to do with linkages between published/submitted data. So it's a way of automating going through the huge swaths of primary research to distill out potential avenues of new/continuing research.

IMHO, I don't think that the field is at the point where the function of the brain can be reasonably simulated by computers. We don't know enough. Sure, there was that group that simulated 10 seconds of half a mouse brain working away... but as to whether or not it was computationally accurate and complete, I'm highly skeptical. I'm sure the researchers who did the actual work are just as skeptical, too. It's the damn journalists.

If you're interested in simulation, people who use NEURON should be able to provide the most realistic advice.

The term "bioinformatics" sounds sexy, but it's the application that gets you the funding. Sorry, don't know enough about that aspect of the field to say more. The same thing with "proteomics." I guess it depends on what you want to do - be a principle investigator, go on tenure track, and try to secure funding to do your own research - or work for a a private company. Having experience in areas with "sexy" will help you get a job working for someone else. Unless you break ground in these sexy areas, getting funding to do basic research may be a little more difficult.

---

Yes, definitely try to join a lab as an undergrad assistant or better yet, get a research-training assignment.

Knowing computers and programing can really help you, but perhaps not directly. Do you want to do neuroscience foremost or computer science foremost? If you're interested in neuroscience, you can work the comp sci aspect into your project. If you're interested in comp sci, maybe join a comp sci group but focus on researching applications for neuroscience.

Feel free to email me if you want to discuss something specific. =)
posted by porpoise at 8:19 PM on November 4, 2007

I'd suggest you learn a programming language, just to verify that this is an appropriate goal for you. An admissions committee may want to see that as well.
posted by sebastienbailard at 9:02 PM on November 4, 2007

UCSD's Computational Neurobiology program accepts people from all sorts of backgrounds. Email or call them about getting a tour or just for information about what sorts of research you can do, since you're adverse to labs.

San Diego is basically the neuroscience capital in the US because you have UCSD, The Salks Institute, and the NSI, all of which are excellent research institutes. They tend to collaborate quite a bit, too. I believe there's a healthy amount of collaborative research between the San Diego institutes and the ivy leagues (namely CMU and MIT as mentioned above).
posted by spiderskull at 10:08 PM on November 4, 2007

You might want to contact this group at Washington University. They have an interesting engineering perspective on neuroscience.
posted by cosmac at 12:37 AM on November 5, 2007

The University of Florida has been doing some seriously interesting stuff with brain/computer interfaces.
posted by saladin at 6:59 AM on November 5, 2007

(echoing what porpoise said) Are you trying to decide about methods or subject matter?

If you want to avoid bench work, there's a lot of computer science embedded in a great many neuroscience sub-areas. I worked for several years in a cognitive neuroscience lab, where we did a lot of fMRI and EEG imaging of the brain-in-action. It's pretty computationally intensive, and there are some interesting programming challenges out there, ranging from visualization and analysis to modeling.

I'd second (third? fourth?) the recommendation to spend some time working in a lab before applying to graduate school.
posted by tew at 8:57 AM on November 5, 2007