A backstage entrance into the world of neuroscience and academia.

Democratisation of neuroscience education

Credits: image by Mohamed Hassan from Pixabay

Over the summer I had a pleasure of participating in one of the most inspiring ventures in academia. I took part in my first summer school during the first year of my PhD journey. It has also been the first year after lockdowns, so one could freely travel and scientific meetings could finally happen in person. Hence, whenever I mentioned to someone that I would be attending a summer school I experienced the following dialogue:

“Awesome! Are you going somewhere exciting?”

“It’s online.”


But doing a summer school from home has its pros. It’s easier to fit in your schedule and to organise, as you don’t have to plan the travel, book planes etc. Because of all that, it is also cheaper and more accessible to everyone. And accessibility is an important keyword for the Neuromatch Academy (NMA), the organiser of the computational neuroscience summer school.

Democratisation of science education

The mission statement of the Neuromatch, Inc., a nonprofit organisation, is the following:

Our mission is to democratize neuroscience.

Neuromatch Inc.

democratisation… hard word. What does it mean? Let’s read further.

Democratization is the act of making something accessible to everyone. Neuromatch seeks to make neuroscience accessible to all by allowing all people from all backgrounds equal access to education and networking opportunities, and providing everybody an equal voice in scientific contribution.

Neuromatch Inc.

Very wholesome! Is it achievable? I took down some numbers that were presented to us during the closing ceremony on Crowdcast. In 2022 NMA ran 2 summer school simultaneously: Computational Neuroscience and Deep Learning. For both courses altogether, there were 4123 applicants from 105 countries and 95% of them were accepted! Male to female ratio of 66% to 30% (the remaining 4% were non-binary or preferred not to say) leaves room for improvement, however, it’s a relatively good balance, given a general issue with that in computational fields. Participants represented many different walks of life and stages of career/education. I think this shows the incredible reach and inclusiveness. How is such a large-scale enterprise organised?

Hello world of computational neuroscience!

This outrageous number of participants is split into groups of circa 10 participants, so-called “pods”, based on a long questionnaire everyone had to fill in during the application process. Each pod has its forum on Discord and a teaching assistant (TA) who was our main point of contact with NMA. There were also general channels on Discord where we could meet and discuss with other attendees from outside of our pod. Everyday we had to watch an introductory lecture and then meet on Zoom with our TA to go together through the problem sheets of the day. The problem sheets were in a form of Jupyter Notebooks (interactive programming platform, where blocks of text are intermingled with blocks of code) that could be run on the Google Colab server, meaning you didn’t have to download anything and everyone could easily work from the browser regardless of the OS they were using (yay accessibility!).

The whole curriculum was excellently prepared. Kudos for the huge collaborative effort that went into creating this. Each problem was preceded with a short explanatory video made by one of the leading scientists and followed by a written description and a block of “fill-in-the-blank” code. The course requires some basic skills in maths and programming and if you are not sure what “basic” means, you are provided with a refresher of all the basic concepts in math, Python and neurobiology to ensure everyone has the same base before the course starts. Beware NMA Computational Neuroscience is not a programming course nor a maths course. It is exactly as the name says – a computational neuroscience course. This means it brings together understanding of mathematical concepts and their application in neurobiological problems. If you are a more advanced student, you can dig deeper in the content and do the bonus sections. If you are a newbie, the whole curriculum might be overwhelming, but it gives you a great overview of computational methods in neuroscience.

All the teaching materials are freely available the whole time for anyone and are accustomed for a self-paced online course. If so, then why bother signing up for the summer school? On top of the theory and exercises, we had to choose as a group a dataset and define a project, so we could test the things we were learning in real time and practice soft skills so much needed nowadays – remote teamwork. We also had evening Q&A sessions with the NMA organisers and other representatives from academia and industry about careers and problems in science. But the unique thing about the Neuromatch were the group tutorial sessions.

Like 10 peas in a pod

First session. Everyone watched the introductory video to the problem we will be studying now. One person is selected to share their screen and solve the task. There is a question; TA’s reply: “does anyone have an answer to this question?”. A moment of awkward silence. Finally someone unmutes themselves and attempts an answer. TA: “Does that answer your question?” A nod from the person asking. Ok, we can move forward. Next person takes over screen sharing and tries to solve the next task. “Does anyone have questions?” Someone shyly asks. Same story. “Does anyone have an answer to this question…?” What a nice job being an NMA tutor… I think to myself.

But after a couple of days any shyness went away, we were freely asking questions whenever something was unclear to anyone and everyone was participating in the discussions. Everyone had equal chance to ask a question and explain (or verify) their way of reasoning to others. The TA was there to moderate the discussion – making sure we are not going off topic, keeping track of time and eventually guiding us to the answer if we were getting stuck.

I must admit such approach to education was a novelty to me. I was used to working on a project in a group or doing exercises on your own and then sharing a solution with a class. But not solving the tasks in real time whilst discussing with others and looking for a solution together. Surprisingly, it turned out to be a very efficient way of learning.

The whole group thing would not have worked if it wasn’t for the awesome people I had in my pod! In a magic way we all complemented each other perfectly. We were all on a similar intellectual level, but with somewhat different background. That ensured fruitful discussion and also a level of psychological comfort – I didn’t feel more stupid than others (neither more clever), which is not an uncommon feeling in academia and I felt that I could contribute. Due to varied expertise in the group everyone could bring something novel and valuable to the discussion. We were learning from each other the whole time.

Better than Tinder?

This magical matching was performed by the clever NMA algorithm, which only adds to the credibility of the organisers – at the end of the day they are supposed to teach you computational methods. The story of the Neuromatch algorithm started in the lab of Konrad Kording, where one of his students, Titipat Achakulvisut, wanted to optimise scientific collaboration. He worked on an algorithm to match scientists based on the abstracts they sent to a conference in order to organise a kind of a speed-dating event. An idea to do it on a larger scale online was accelerated by the global pandemic of Covid-19. Neuromatch successfully run their first online conference and later in 2020 they launched the summer school.

The algorithm clusters applicants with minimal distance between them computed based on the answers provided in the application process. The goal is to match people with similar interests, same preferred language and timezone (remember, it’s a global initiative!). You can read more about it here. As one of the NMA organisers, Gunnar Blohm, mentions in his article, the algorithm required manual tweaking in the first two runs of the NMA summer school. However, it is constantly being improved and one can follow the latest developments on Titipat’s GitHub.

Nobody’s perfect

As good as it may all sound, the course was not 100% perfect. Each pod had project tutor who supervised us daily and we were also supposed to have a project mentor, a senior scientist that we could meet twice to get some advice from an expert. My group didn’t have a project mentor at first. Our tutor organised someone, but it turned out they did not have the relevant experience for our project. And they did not show up to a subsequent meeting. However, it did not impact our project at all. Given the massive scale of the whole initiative, such minor hiccup does not disturb the big picture – the whole organisation is still very impressive!


The fact that Neuromatch works out the way it does and is so successful is very inspiring. As they paved the way with computational neuroscience, more subject matters are on the way! Hopefully soon we will bridge the educational gap between the nations and we could collaborate on solving the scientific issues as a global community more easily and efficiently. If this is a legacy of Covid – at least we have one positive thing that came out of the pandemic. 🙂