home projects sketches open source talks

Problems with the Human Brain Project

24 October 2015

or how I’d spend 1 billion euros to accelerate our knowledge of the brain

The Human Brain Project is a project put together by a bunch of scientists in Europe to create a whole human brain simulation with 1 billion euros ($1.1 billion US) pledged to it over 10 years. The goal is very grand and noble, but I’m afraid will not achieve its results. In July of 2015, ~800 neuroscientists sent an open letter to the project questioning the validity of the project. Just because 800 scientists don’t agree doesn’t mean the human brain project is wrong, but there are strong concerns. A model of anything is almost always wrong, check out All models are wrong but some are useful. 1 billion euros and 10 years is a lot of money and time to through at an experiment with no clearly defined goals. To test the simulation, the whole brain simulation needs to be built and running first. What happens when all that money and time is spent and then they get to testing it and find out that there are a bunch of mistakes and parts that need to be redone. How do you know what the right parts are to model? Do you model neurons? Glial cells? neurotransmitters? What about DNA and protein? There are a lot of different parts of the brain to think about when putting together a model, this article does a good job explaining the problems with Human Brain Project.

on the brain and brain models

We know so little about the brain because it is extremely hard to study and so brain models are one potential way we can test out our ideas. Building models and simulations around the brain is a topic dear to my heart. I’ve spent countless weekends reading papers, examining open source brain simulations, and contemplating different brain models I’d like to test that might provide us some insights into how the brain works. I do believe that building brain models will eventually lead us to important discoveries about the brain and so I do understand the reasoning behind the Human Brain Project, but it is just not realistic now to build models without more understanding of how the brain works. I hope to one day spend time building some simulations to test out some of my ideas.

C. elegans

When studying the brain and science in general, we often study other organisms. We call these model organisms, which are organisms that we study extensively because we expect the learnings to transfer to other organisms, especially humans. For example we study monkey brains because they are one of our closest relatives. One organism that many brain scientists study and consider a model organism is Caenorhabditis elegans, or C. elegans for short. It is a roundworm that is approximately 1 millimeter long and extensively studied(~1000 new papers/year). The human brain has about 10 billion neurons, while C. elegans has EXACTLY 302 neurons. We have basically no idea how C. elegans and its neurons work. How does its 302 neurons allow it to feel, function, find food,etc? If we can’t understand such a tiny organism, how do we ever expect to understand a human brain with 10 billion neurons?! Anyway, now to how I would spend the money.

I would built new hardware tools to study C. elegans in finer detail. Could we build a tool to wire up every single neuron so that we can watch what happens to every single neuron when different stimuli is applied to C. elegans. This tool would basically be a chamber or lab on a chip for the worm to live in and we would inject different stimuli,food, etc to observe it. Could we add cameras,chemical, and electric measurement instrumentation to capture the state of every neuron as we run experiments on the worm. It is almost impossible to do this now because the worm is already only 1 millimeter in length and the instrumentation to wire up each neuron must be even smaller.

After building this new generation of tools to study C. elegans. I would either research how to bring the costs of the tools down or subsidize the tools so that EVERY SINGLE school (high school and up) and classroom can do these studies. I would create a package to send to each classroom that consisted of 10 chambers, classroom materials for studying, a teacher’s guide, and an initial batch of the worms.

We need to get more people interested in studying the brain and coming up with interesting experiments. Having access to run tests the same way cutting edge neuroscience research is done is a great way to do that. With all the research and learnings resulting from these tools, I would then start applying the findings to building a simulator for C. elegans. I would want to make sure the simulator models the correct aspects of C. elegan’s neurons so that the tests on the real thing match the tests on the simulator. There already happens to be a fully simulated C. elegans organism being built in an open source way called openworm. If we are ever successful in building these kinds of tools needed to study C. elegans, then I would go to another animal, probably the shrew as it is the mammal with the smallest brain. Along the way I’m sure there would be lots of learnings that would help us understand ourselves better.