I am graduating this year with a Dual Degree from the Electrical Engineering Department, specializing in Communication and Signal Processing. I am joining the Bionet lab in the Electrical Engineering Department at Columbia University to start my Ph.D. in Computational Neuroscience. During my time there, I will be working with Prof. Aurel Lazar on uncovering the functional logic of the fruit fly brain.
My journey through insti has been a meandering one. I started as a B.Tech student in Engineering Physics, switched to a B.Tech in Electrical Engineering, and then switched again to a Dual Degree in Electrical Engineering. My research interests have also evolved substantially during this time, with the roundabout route towards my current research interests helping to reaffirm my penchant for interdisciplinary research. This article will focus on interdisciplinary research for a few important reasons:
- The world is rapidly becoming incredibly complex and interdisciplinary research is exploding
- Yet, by its very nature, it is initially harder to navigate this larger space, making advice all the more pertinent
- And because this is what I am most qualified to write about.
Computational Neuroscience is a quintessentially interdisciplinary field: as it attempts to mathematically model the biological underpinnings of human cognition, it naturally draws ideas from and enriches a variety of other fields that leverage human intellect including Physics, Biology, Statistics, Computer Science, Engineering, Psychology, and Philosophy. But this kind of broad scope is indeed overwhelming, causing a lot of confusion about how to get started and about formulating a coherent yet substantive narrative for a grad school application.
This brings me to my first and most important piece of advice: do not view your goal in an interdisciplinary field as becoming a jack of all trades. The goal is not to have a perfunctory understanding of many approaches to tackling a given multi-faceted problem. Rather it is to master one approach so that you become a valuable member of a team attempting to construct a multi-pronged solution. What this translates to in practice is to find one or two fields of mathematics or engineering that you like, and to then see how they can be applied to the field of your interest.
In my case, I love the field of Statistical Signal Processing. I thus focused on developing a skillset around this niche and then asked how I can apply these tools to Neuroscience. This vastly streamlines the learning process and makes it that much more interesting, helps build intuition for what approaches work and don't work, and also allows you to pick your electives based on a well reasoned bigger picture (thus answering for yourself the question of *why* you are subjecting yourself to the drudgery that is a part of most courses).
Having a strong niche is also extremely useful for a Ph.D. application as it shows that you have the tenacity to wrestle with and master ideas and that you have an important perspective to add to the lab right off the bat. Picking such a niche also makes it easier to search for potential Ph.D. advisors (which is highly non-trivial and is something that you should start doing many months before the application deadlines). Using research internships and research in insti to secure this niche, while using courses and course projects to cultivate a broader understanding of the field is an effective two-pronged strategy to keep in mind.
My second piece of advice is related to the first: the goal of applying to grad school is getting your foot in the door, so put your strongest foot forward. What this means is that apart from picking a niche to focus on, you should also be applying to groups that will understand and value your academic experience so far. The clearest application of this piece of advice comes in picking which department to apply to. So for me, as a EE student interested in Neuroscience the dilemma was between applying to Neuroscience and EE departments. I hedged my bets and ended up applying to a mix of EE and Neuro departments. I got into a healthy fraction of the EE departments that I applied to, while I received interview calls from some of the Neuroscience departments but was not finally accepted to any of them.
Without any knowledge about the other applicants, the one important reason for this outcome that I can certainly point out is: final decisions are often made by a department-level committee that comprises interested advisors along with other professors. So while many of the Neuroscience departments "highly encourage" applicants from other departments, my background (research and coursework) is difficult to assess for the other members of the committee. On the same note, recommendation letters from EE professors won't hold as much weight in a broadly focussed Neuroscience department. Subsequent email inquiries with interested advisors in these departments also provided more credence to this theory. So while I was fortunate enough to be accepted to my dream lab, I still learned an important lesson here in strategic positioning.
Based on my apping experience I would thus strongly recommend you to try and apply to your home department for the most part. This might seem like denying yourself a fresh start (if you desire one) but something that is well known but is often forgotten in the midst of actually applying is that grad school is a lot more flexible than undergrad. So it does make sense to base your camp in home territory and then set off to engage with camps of researchers in other departments of interest.
For anyone struggling to choose between a "hot"/mainstream field and an offbeat field, there are some tradeoffs that you should consider very carefully. As a new researcher, the mainstream field is lower risk as there are much more resources (mentors, co-researchers, software/tools) available to help you get started. However, by the same token, such a field is always prone to overcrowding (at the beginner/intermediate level). The offbeat field on the other hand is sparsely populated, but by the same token may also have a smaller support ecosystem. This willingness to handle competition or newbie trouble must then be weighed against the payoffs, both short-term and long-term.
In the short term, the mainstream field is much more likely to yield "higher" rewards, if evaluated based on the number of publications. However, the question for the long-run is whether you are too late to make a fundamental impact in this already flourishing field. Much like how the Newtonian laws of motion can only be discovered once, so it is with ideas in every field. So you have a choice to make: will you reap the incremental yet stable rewards of being a late adopter of a currently flourishing field, or would you like to shoot your riskier shot at trying to lay the foundations for *potential* future explosive growth in a currently offbeat field? Thus there is a standard risk-reward trade-off curve that you must estimate and use to make a decision. As with most things in life, things are rarely binary and your final decision will lie somewhere on a spectrum, and there are of course a myriad host of other factors to consider while picking a research direction. And indeed, this article is in no way meant to disparage "hot" fields, only to encourage you to think carefully about your choices.
In my case, I am drawn to Computational Neuroscience because it is a field in which the fundamental building blocks are only just being discovered and so there is an opportunity and need for the kind of foundational research that excites me the most. Ironically, while computational modeling in neuroscience was fairly offbeat a decade ago in the broader space of neuroscience, it is rapidly carving out an ever-larger niche for itself, fuelled by explosive growth in "mainstream" experimental neuroscience. However, this broader adoption is yet to diffuse to the shores of insti as of this writing. As a result, for better or worse, I had to be very independent in finding relevant coursework and intersectional research problems during my undergrad.
I would now like to dole out some generally applicable apping advice: The best school for you may not be the best-ranked one because rankings are only (quantitatively defined) subjective, low-dimensional, macro-indicators. For a Ph.D., the most important thing is who your advisor is going to be, followed by what the overall lab culture is like. What department and university this lab happens to be situated in is but a detail *if* you intend to continue in the same field. If on the other hand, you wish to radically change your career path after grad school, then the network effects of a higher-ranked institute come into play. Thus, ceteris paribus, it is good to aim for high ranked universities. However, you should also be flexible about this and factor in your personal career goals. On a related note, famous researchers aren't necessarily the best advisors. While said researcher may be brilliant at research, responding to the needs of a fledgling Ph.D. student is an entirely different ball game. So talking to other students in the lab is crucial before deciding to apply, and for subsequently accepting an offer.
This then brings us to the question of shortlisting schools to apply to. It is generally advisable to target 7-10 schools and to shortlist them as early as possible. The internet is your friend in this hunt, and I highly advise making a hyperlinked sheet of potential advisors well in advance of essay deadlines. It is highly recommended to hedge your bets by finding at least 3 Professors that you are interested in working within each of the departments that you apply to. Note that sending emails beforehand to determine if a particular Professor is accepting students this year does not often work. This makes it all the more important to find more than one potential advisor in the department.
The application process itself is extremely taxing, and customizing your statement of purpose (SOP) for each department is time-consuming and the application fees are also high. These are both strong reasons to not apply to more than 10 schools as a rule of thumb. A tried and tested piece of advice for writing the SOP is to write the first draft in free flow, and to subsequently refine it over multiple iterations. It can then be cosmetically edited to fit the needs of each university application. The GRE and TOEFL are not very difficult exams, but it is best to prepare a month or two in advance. I found the official guides along with Magoosh's resources to be more than adequate for preparation. Finally, except for one interview (which was around 2.5 hours long), all my Ph.D. interviews were brief (a little over half an hour), largely informal, and did not require any extra preparation per se.
I hope this article proves useful to some of you.
Good luck!