For the last eight years, I called myself a cognitive neuroscientist. Throughout undergrad and grad school, I spent my days finding out how humans make decisions based on the information they extract from the outside world, and what factors play a role in determining our choices.
Half a year ago I started working with mice, asking many of the same questions. I was (and am) excited about the possibilities of recording from actual cells. I was (and am) inspired by the many genetic tools available, and the opportunities for collecting quantities of data from single individuals that are very rare in human subjects research. I also thought it would be cool to do something different – wrap my head around some fresh ideas, and perhaps build up a useful combined skill set.
“Science would be ruined if (like sports) it were to put competition above everything else, and if it were to clarify the rules of competition by withdrawing entirely into narrowly defined specialties. The rare scholars who are nomads-by-choice are essential to the intellectual welfare of the settled disciplines.”
– Benoit Mandelbrot
Without wanting to sound as grandiose as Mandelbrot, the transition has been very interesting so far. There were some obvious obstacles and challenges: I’ve learned how to do surgery, and I’m only now starting to understand what a Cre-mouse really is and why you’d want to combine it with something I’d put on a bagel. More interestingly, several things have surprised me in the transition.
The most immediate change was responsibility towards study subjects, and the time involved in data collection. Laboratory animal husbandry is a large and complex field, with many (very necessary and important) regulations, checks and protocols that need to be followed daily. The impact on my daily schedule has been pretty large (especially compared to the last year of my PhD, when I was mainly analyzing data and writing papers): you can’t just pack up for the weekend without taking care of your study subjects, or work from home during a sick day without making arrangements. Along with other changes to the projects I’m working on, this has made long stretches of uninterrupted working time more scarce.
Working with animals has also made it much more difficult to talk to people about my research, especially when I don’t know them well. My former colleague Ashley Juavinett has written about this beautifully; I consider myself pretty open, but I still have a lot to learn about best ways to have these conversations.
One of the most striking differences between human and rodent neuroscience is the extend to which people build their own tools. Human neuroscience, and especially psychology, uses a lot of techniques that are pretty established and mostly outsourced to commercial vendors. There is a large overlap with consumer (behavioral testing, eyetracking, VR) and medical (EEG, MRI) technology. Researchers with a psychology background are usually not great engineers, so most tools used in human neuroscience are bought rather than built. Due to safety regulations for neuroimaging, those tools are mostly ran by dedicated support staff. In systems neuroscience, on the other hand, people build their own microscopes, design 3D printed custom parts, and I’ve heard people solder their own electrodes.
This emphasis on tool-building has several consequences; some great, and some less so. On the good side, it’s been absolutely amazing to see the creativity with which systems neuroscientists design their setups. They have a much deeper understanding of the tools they use, which allows for a lot of flexibility in experimental design. Mice run on balls or through mazes, swim in water, navigate through virtual corridors, spin lego-wheels, grab handles, lick spouts and run for cover. By contrast, human study participants usually press buttons on a keyboard, or maybe use a joystick when the experimenter is feeling adventurous. The larger range of experimental setups in rodents has taught us a lot about for example the relationship between running and the processing of visual information, which would be harder in large species that have to lie still in a scanner.
However, this creativity also comes at the cost of reusability and standardization. Each project has its own unique setup, which sits unused when the main experimenter isn’t collecting data (or when a project is finished). It can also be very tricky to exactly document a setup with all its idiosyncrasies, and to replicate an experiment between labs – what if the material you used to sound-insulate your rig actually matters? The insane amount of customization that goes into designing and debugging a setup is practically impossible to capture in a paper’s methods section.
The obvious individual cost is an skills-time trade-off: projects take much longer to complete in systems as compared to cognitive (heard in the lab: “the first four years of grad school are for building a microscope”). Whether this is worth it may be a very individual balance; while it can be very satisfying to build things, I’m personally not convinced that reinventing and designing the same tools is the best use of most scientists’ time.
I’m very curious to see where things will move in the next years. With more labs starting to address standardization and reproducibility [full disclaimer, I work on the IBL], will systems neuroscientists grow to rely less on their DIY skills?