PLOS Science Wednesday: Hi Reddit, we’re Eric and Konrad, and our
article in PLOS CompBio reveals flaws in the typical analysis methods
used in neuroscience, and suggests improvements for the research
community moving forward – Ask Us Anything!
Hi Reddit! We are Eric Jonas (a postdoctoral researcher at UC Berkeley
in Electrical Engineering and Computer Science) and Konrad Kording
(Professor at Northwestern and RIC). Our research focuses on trying to
understand how the neurons in the brain compute and give rise to
behavior. A lot of what we do is come up with mathematical techniques to
understand neural data and work on new methods to acquire this data. We
recently published a paper titled “Could a neuroscientist understand a
microprocessor?” that tried to address if the analysis methods we
frequently use in neuroscience are likely to provide the level of
“understanding” we seek. One of the biggest challenges facing
neuroscience is that we don’t actually know ahead of time how neural
systems work, so validating analysis techniques can be a catch-22. We
attempted to use these techniques on a microprocessor – a system we
understand really well – to see if we could make sense of how it works.
This ended up being quite difficult, and we suggest ways that we might
move forward as a community to make sure our analysis methods really do
what we hope they will. Our study has also been written up in the
popular press. Read the articles in Arstechnica and The Economist) to
learn more. We will be answering your questions at 1pm ET – Ask Us
Anything! Follow Eric on Twitter @stochastician Konrad at @kordinglab.