Most posts here are electrical or mechanical, with a few scattered hacks from other fields. Those who also keep up with advances in biomedical research may have noticed certain areas are starting to parallel the electronics we know. [Dr. Rajib Shubert] is in one such field, and picked up on the commonality as well. He thought it’d be interesting to bridge the two worlds by explaining his research using analogies familiar to the Hackaday audience. (Video also embedded below.)
He laid the foundation with a little background, establishing that we’ve been able to see individual static neurons for a while via microscope slides and such, and we’ve been able to see activity of the whole living brain via functional MRI. These methods gradually improved our understanding of neurons, and advances within the past few years have reached an intersection of those two points: [Dr. Shubert] and colleagues now have tools to peer inside a functional brain, teasing out how it works one neuron at a time.
[Dr. Shubert]’s talk makes analogies to electronics hardware, but we can also make a software analogy treating the brain as a highly optimized (and/or obfuscated) piece of code. Virus stamping a single cell under this analogy is like isolating a single function, seeing who calls it, and who it calls. This pairs well with optogenetics techniques, which can be seen as like modifying a function to see how it affects results in real time. It certainly puts a different meaning on the phrase “working with live code”!
Afterwards there was both a lively Q&A session and multiple small group discussions. Unlike high energy physicists and their multi-billion dollar experimental facilities, [Dr. Shubert] believes this technology will be accessible to modest research labs in the near future. This potential both excited and worried the audience, as individuals cited various pieces of science fiction that loomed close to become our reality. There is agreement that public policy lags behind technology development, but no consensus on how that would be addressed.
And a final note of trivia: with better understanding of neurons, [Dr. Shubert] says we now know real neural networks to be quite different from the software creations they inspired, spurring research into a new generation of software. This and more fascinating insights absent from the video stream are great reasons to come to Hackday Los Angeles meetups in person!
Single neurons are really not at all like function calls.
Virtually everything in the brain works on highly redundant population coding. It’s obvious why – losing one or two cells would be catastrophic if it didn’t, just like losing one or two transistors in a processor, or yes corruption of critical code paths is catastrophic.
Looking at a single neuron is great, but your insight is absurdly limited. Without knowing what the population encoding is doing, the same activity could mean a hundred different things. It’s like looking at a single memory location on the stack. Maybe it’s a single number, maybe it’s a pointer, maybe it’s actually on the heap and you’ve got a single byte of a huge object, or maybe it’s unallocated and just garbage from the last time it was used. You can get some insight by seeing how often it changes, but… not that much.
There’s also the classic paper http://journals.plos.org/ploscompbiol/article?id=10.1371/journal.pcbi.1005268 which includes the quote “We can thus conclude [these gates] are uniquely necessary for the game—perhaps there is a Donkey Kong transistor or a Space Invaders transistor.”
Along with “Can biologists fix a Radio?” It’s one of my favourite papers.
I have a neuro/bio degree and 3/4 of an EE degree, and I think the EE degree is more useful for studying brains.
Fun paper. Thanks for the link.
I remember hearing-correct me if I’m wrong-that our current deep neural network tech was made after reverse enineering the visual image processing part of the brain. This was made simpler than what is being proposed because image processing is massively parallel, and could be analised by looking at the structure of the net as a whole.
@Olsen – Interesting! Thanks for sharing. If anyone has a reference or more info about that – I’d be interested
Ish. Convnets are more-or-less analogous to the visual system. Other networks designs, not so much.
“And a final note of trivia: with better understanding of neurons, [Dr. Shubert] says we now know real neural networks to be quite different from the software creations they inspired, spurring research into a new generation of software.”
A better engineered spherical neuron.*
*From a cow.
Wow, seems risky to me in regards to worst case scenarios for body and mind control as well as disease causing forensically clean. Reminds me of nanobots or really magnetically guided/controlled viral or pathogen vectoring to cause whatever disease or psychological or even body control manipulation… and we know how great the healthcareless and acts like an attorney racket industry can be in many places.
Not sure there is a reason to see how the neuron pathway function works using this method other than seems the gaps in cell formation into organs?
Kind of like radio isotopic labeling if neurons or something when there are better ways like using SQUID’s or other forms of spectroscopy, tomography or holography for issues… though I guess like for more detailed not requiring invasive morbid surgeries say like for epilepsy and targeting out of control cell growth to remove remotely. Maybe benefits for other healthcare issues.
Neat genetic engineering and methods though. Cellular reprogramming is interesting also. Just think… men can reproduce with other men since men have X and Y chromosomes. Really interesting. I want to see the OpenWater Company holography methods so can have more intricate 4D images with less instrumentation.
I wonder if the materials can also be some sort of acousto optical tunable filter (AOTF) like material? I had tested the feasibility of a Brimrose NIR spectrometer that used a Tellurium Dioxide crystal to tune the NIR signal via the acoustic signal. http://fluoview.magnet.fsu.edu/theory/aotfintro.html
I wonder if something similar is done to tune emissions in other frequency ranges to study the most feasible frequencies that are most sensitive the a minimal transmission to activate or other utility in safer ways?