What is the discussion?
Lex Kravitz and Mark Laubach will host the Hack Chat on Wednesday, February 19, 2020 at noon Pacific Time. Time zones got you down? Here's a handy time converter! Unless, of course, you look inward to the most wonderfully complex structure ever found: the brain.
Open source tools and neuroscience, and the OpenBehavior project
Can everyone introduce themselves please
Openbehavior.org, in 2016, to further open-source methods in behavioral research. In the intervening years, it seems that many people in the scientific community have started using and publishing open-source methods, so we’re here to share a few interesting projects with this community, and also answer any questions about neuroscience, research, life, etc!Sure, are we good to get started? I'll kick off with a bit of an intro: My name is Lex Kravitz and I am a professor in the Psychiatry department at Washington University in St Louis. With our co-host, Mark Laubach, we started a website,
Im A Essa (discontinued Phd in AI) and software engineering bsc
I'm Mark Laubach. I am a neurobiologist, with lab at American University in DC. We study how decisions are made in the brain, and use a lot of custom made tech in our research.
Yes, please, let's start. And welcome Mark and Lex!
Thanks! I thought we'd start by sharing our website and highlight some of the more interesting projects that we've covered on it
Camera Control is an open-source software package written by postdoctoral fellow Gary Kane that allows video to be recorded in sync with behavior. The python GUI and scripts allows investigators to record from multiple imaging source camera feeds with associated timestamps for each frame.
openbehavior.org, with the goal of promoting open-source methods in neuroscience. It's something Mark and I do for fun, along with the students in his lab, mainly Sam White and Linda Amarante. Maybe we can start with some youtube videos to demo some cool recent advances in open source neuroscience?We made a website,
One of my favorite open-source tools (that has been hugely influential in behavioral neuroscience over the last year) is DeepLabCut:
This is a markerless pose-tracking package, that has been used to improve video tracking of mice, flies, and *many* other laboratory applications
That's pretty cool
Wow, that's exactly what I tried to design in my old job 20+ years ago! We wanted to track where a rat was pointing his nose. Machine vision just wasn't up to the task at the time
@Dan Maloney That was my Master's project in 1990! Did not work so well.
Cool! yeah most methods before this used "markers" - a piece of tape or something like that
Perhaps you can define the difference between random movement, purposeful movement and a behavior?
how did you solve that? training etc?
Intro for me (Charlie L aka CyberchuckTX)
: lifelong geek/nerd/programmer interested in these topics.
openbehavior.org" times out when I try to access it ...BTW, "
http://www.mousemotorlab.org/team) put a lot of effort into making this tool accessible - notebooks, demos, etc. As a result they are one of the best known solutions for this and have gained a large user base - open source at work :)Beyond being an amazing application, the team that made this (
Chuck that sounds like a philosophy hackchat :)
All this is all good, but what are you trying to achieve with all of this?
I'll bite though - no behavior is random, and all behavior is "behavior". Even blinking or breathing. With animals it's very hard to ascribe purpose, so I tend to try to just quantify the behavior and leave it at that - was there more running, less running, more pellet eating, etc.
so happy to see open source tools for neuroscience being highlighted here!
intro: I work at the university of Sussex and have been advocating for the use of open source tools for neurosciences for almost a decade now (almost as long as Iṽe been in neurosciences :P )
Hi there! Thanks Lex and Mark for sharing your time with us. I've been checking out your projects, and find your field of work incredibly interesting. I was wondering... How different is electronic equipment used to analyze the spikes of different electrode interfaces? (EEG vs ECG vs EMG... be it through implants, skin contact...) Do they all use similar combinations of amplifiers+filters+DAC, and differ only at the physical connection with the body? Or do they require different designs/methodologies for data adquisition and processing?
Hi Andre!!! Yes I know your work thanks for stopping in!!!
@Essaamar and others: Maybe this will help...
As a consequence of making tools and running OpenBehavior, we also wrote an article about open source tools and the need to make it easy to publish them and track their use. Check it out:
here is some work some communities are doing tthat is related to neuro/openhardware
Open source hardware project from the department of Neurosciences at Sussex University
@Lex Kravitz Thanks! also thanks for putting together open behavior!
@Mark Laubach ! Thanks to you too for putting these things out there :)hi
Let's just say that I have a goat that I want to motivate to eat this and not that.
@Andre Maia Chagas. Special thanks is also due to Samantha White and Linda Amarante, who have written most of the posts and posted the Tweets.Thanks
@Juan Carlos Jimenez The electrical interface can be very different on different tissues, so the electrodes themselves are very specialized - the wires we use for recording single neurons are much smaller and higher impedance than leads for recording EEG for instance. Often the amplifiers are also specialized to get the best signal:noise from the specific application. As one concrete example, recording single neurons requires sampling rates of ~40kHz, while for recording brain waves via EEG it can be <1kHz. But after the data is stored on a computer it's a lot of the same approaches you might use to analyze any electrical signal - filtering, identifying waveforms, looking at the frequency or amplitude of different events
Do you guys get a lot of projects on open behavior that is not from people in universities?
@Andre Maia Chagas almost none actually... I think it would be a huge benefit to find ways to build bridges between academic scientists and non-academic scientists or citizen scientists
ok, I wish we could come up with a way to make this academic barrier smaller/non-existent. There is so much knowledge in non-academic communities, I think everyone ends up losing by not having closer collaborations.
I'm hoping to be able to do some DIY work for a dear friend of mine who recently had a terrible accident that rendered him (mostly) paraplegic. He's an engineer by trade so he's vitally interested in anything that could help him interface to the outside world.
For us, the challenge is to get researchers to live by this principle: Don't let tools drive your research questions. Make your own tools to solve the problems that you are interested in. This applies to research labs as well as teaching and many other endeavors.
@Charlie Lindahl I'm really sorry to hear that
@Charlie Lindahl sorry to hear that... Don't know if it was already mentioned, but look into openbci and openbionics ()
@Charlie Lindahl Sorry also. There is a field called Brain-Computer Interfaces that is intended to address those kinds of issues. Unfortunately, most of it is not open source.
@Andre Maia Chagas is spot on with those projects.And
I left research on what was called BMIs at the time due to closed thinking in the early research community. Closed and competitive.
Brain Machine Interface
Brain machine interface?
Yes some scientist has already hacked the brain
Closed and competitive? What does that mean?
idea was to hook up neurons to control devices
@Mark Laubach do I guess correcty that most of the magic is in the signal evaluation, while the collection of signals for control of the mechatronics is rather routine?
no sharing of results, methods, code; venture firms offering lunch
@Thomas Shaddack Signal processing and algorithms for decoding and control were/are not so easy given noisy neurons; biggest problem then, and maybe still now, was keeping electrodes viable long term
Brain machine interface sounds a bit over the top if you have the nervus vagus
@Thomas Yes. The older work was targeting motor cortex.
well, I wouldn't go higher than the brain stem
@Thomas and the fact that EEG can be used
hmmm... that's rather difficult, right?
@Thomas Brain stem is hard because a lot of critical circuits pass through there; surgeries is a serious challenge
I think implants would be best
@Mark Laubach where is the main set of problems with the electrodes? the metal-neuron interface?
true. the motor cortex would actually be safer
There was a group at IBM that was using EEG successfully to control simple circuits like light switches and toilets; simple problems that need to be addressed for independent living
@Mark Laubach i heard about some tricks with conductive polymers, forming around the cells.
@Thomas Shaddack Glia growth limiting isolation of neurons
Getting back to tracking the movement of a rat in a box, is that a behavior? Drinking water from the water tube certainly might be classified as a behavior. how is that useful?
Anyway... we don't really cover BCI/BMI on OpenBehavior
hackaday.io, no?I think there are a couple of eeg/bci projects here on
Yes, I worked on some of the NeuroSky stuff, controlling fursuits. It was very unsuccessful.
It's nowhere near precise enough and everything on the market uses more interpolation than actual data.
@Chuck Glasser : how is that useful in what sense?
@Chuck Glasser Yes, drinking from a tube is a behavior, and understanding the decisions that lead to drinking now or later, or one fluid or another, are a major topic of work in the field
@Chuck Glasser Glad you asked! Yup a rat in a box is a behavior. Our lab uses that to measure activity in obese rodents (our lab studies obesity for anyone who missed the intro). A big question in obesity research is why people with obesity are less active than lean people. We're using movement in a box as one read-out of activity, so we can identify what it is about obesity (or junk food diets) that leads to inactivity.
I worked on an eeg to helicopter before in university, its very noisy
what is your experience with reproducibility of results with OSH compared to quality controlled commercial products? Any feedback from the scientific community on that?
@Lex Kravitz YES! Obesity, decision making, sensorimotor control.... all studied using simple behaviors like drinking fluid from a tube; and one of the most covered topics on OB
@Thomas The nice thing about OSH is that the papers that come out describing tools normally come with calibration protocols and specify equipment working limits, etc
I wish that EEs and technologists in general would come to appreciate the complecity of even the "simplest" of behaviors
so you know quite well what you get/what you can do with it.
https://en.wikipedia.org/wiki/Terminal_nerve (if it is a vetigial element) with an analog opto stimulator. You could have the light receptor just under the skin with a inductor to power the circuit... You could even embebed it in your glasses...why not reuse the terminal nerve
@Thomas not being snarky at all, but are there really quality controlled products in the neuroscience field? My dad did quality control for Ford Motor Co back in the day, and there is nothing in science vendor world like what auto companies did on the assembly line
@riley.august Interesting! Yeah my feeling on the consumer grade EEG products has always been that they are not useful for measuring what they claim (or at least not with the EEG signal, perhaps with other sensors like motion). One reason I'm negative on this is that neuroscientists don't know how to measure these things with research-grade equipment....