First Movie of an Entire Brain's Neuronal Activity

KentuckyFC (1144503) writes "One of the goals of neuroscience is to understand how brains process information and generate appropriate behaviour. A technique that is revolutionising this work is optogenetics--the ability to insert genes into neurons that fluoresce when the neuron is active. That works well on the level of single neurons but the density of neurons in a brain is so high that it has been impossible to tell them apart when they fluoresce. Now researchers have solved this problem and proved it by filming the activity in the entire brain of a nematode worm for the first time and making the video available. Their solution comes in two parts. The first is to ensure that the inserted genes only fluoresce in the nuclei of the neurons. This makes it much easier to tell individual neurons in the brain apart. The second is a new techniques that scans the entire volume of the brain at a rate of 80 frames per second, fast enough to register all the neuronal activity within it. The researchers say their new technique should allow bigger brains to be filmed in the near future, opening up the potential to study how various creatures process information and trigger an appropriate response for the first time."

Comments

Mapping the Nematode?

[LifesABeach]

So many questions.
Could a complete mapping of the neural network be accomplished?
Would it be possible to artificially trigger a neuron to verify the mapping?

Re:Mapping the Nematode?

[Rei]

Why wouldn't it be? Light-sensitive proteins are quite well understood.

All of this leads to a really fascinating possibility down the road...

Part 1: Requirements:

1) Genes are inserted into the nucleus of every neuron.
2) Probes which can receive on one or more optical frequencies** and send directionally on other frequencies (which we'll call A, B, and C) are inserted all throughout the target brain.
3) The genes from #1 flash upon synapse**, allowing the probes in #2 to receive the signals
4) The genes from #1 force a synapse when they receive frequency A from a probe.
5) The genes from #1 suppress synapse when they receive frequency B from a probe.
6) The genes from #1 force the cell to commit apoptosis when they receive frequency C from a probe.

Part 2: For each neuron in the brain (conducted in parallel):

1) The neuron's behavior is studied relative to its neighbors in order to learn precisely what factors control its activation levels. This requires a very accurate neural model, and probably requires a lot more more than a simple one-frequency "I'm firing" signal in #2 and #3 of part 1.
2) The neuron is simulated in a computer based on said inputs
3) The neuron is ordered repressed when the simulator doesn't want it fired, and ordered fired when the simulator wants it fired.
4) The system works its way through all of its neighbors that it influences, doing steps #1-3 of this part upon them and putting them under control of the simulation as well.
5) Once a neuron is entirely isolated and can be handled entirely within the simulation, the signal is sent for apoptosis.
6) This pattern continues until the entire brain exists only in the simulation.

And thus you take any living entity and entirely digitize their consciousness, without any single moment defining their transition from the physical world to the digital one, and without "copying" them.

This is a key first step in something I've been thinking about for a long time, and I'm thrilled to see it. I doubt I'll live to see all the steps, or that anyone alive today will. But I'm thrilled to see the first steps taken down this road.

More near-term, one can envision all sorts of incredible properties with an optical communication link set up with cells. For example, imagine that you instrument cells in a cancerous organ with genes that can be instructed individually to force the cell into apoptosis, and which flash on various frequencies corresponding to various cellular activities. You look for cellular activities which correspond to cancerous behavior, and when you see them, you tell that cell to kill itself. You really have something way better than all of that unrealistic "nanomachine medicine" stuff that sci-fi writers have been obsessing over for ages.

Re:Mapping the Nematode?

[Ferrofluid]

You look for cellular activities which correspond to cancerous behavior, and when you see them, you tell that cell to kill itself

That's kind of what's already supposed to happen naturally inside the human body. Cells are supposed to kill themselves if they are severely malfunctioned or are likely to become cancerous. However, if enough of these fail-safe mechanisms are damages within a cell, then that cell becomes cancerous. That's why cancer is so difficult to treat, and why your own immune system has difficulty attacking it -- the cancer cells have gone rogue and are no longer "following orders" to kill themselves.

So, if you were able to insert genes into cells, which would allow the cells to kill themselves upon activation by a certain light wavelength, then what would happen? Say you illuminate the tumour with that particular wavelength. Perhaps 99.9% of the cells will undergo apoptosis, as instructed. But maybe 0.1% acquired a mutation which disabled your fail-safe genes. Now what? Congratulations -- the cancer has now evolved to be resistant to your light-induced apoptosis commands. And you're back to square one.

80 frames/second

[i kan reed]

If that captures everything, that's the interesting part to me(though I'm sure it's been known to actual neurologists forever). That means the "clock speed" of the human brain is really really really really low, more or less, right? Like our consciousness is pretty much exclusively the result of massive parallelism?

Re:80 frames/second

[i kan reed]

Sure, okay. There's more than a little evolutionary separation since then. Do you think there's a good reason to assume the difference is dramatic?

Re:Where's the Video?

[Anonymous Coward]

Here you go, direct link to the YouTube upload.

Having no idea...

[NMBob]

Do neurons just fire or not fire in a binary fashion, or are there different levels/voltages of on and off?

Re:Having no idea...

[raftpeople]

Short answer: yes, different levels.

Longer answer: There is much complexity and the real answer would probably cover a few books. But a few highlights: neurons have their firing rate and their spike levels modulated by a variety of things in the brain. Glial cells (which are 10x more numerous than neurons) inhibit and disinhibit neurons, communicate with each other and are involved in computation. Some neuron communicate with a continuous flow of protons (inner ear, acceleration detection), some fire locally on their dendrite instead of the typical method of sending a signal down the axon. There have also been recent discoveries linking microtubule quantum vibrations to anesthesia effects (implying it is part of computation).

translated

[Charliemopps]

And we finally know what nematodes are thinking: http://i1.ytimg.com/vi/GpEDsoZ...

Analog. not digital.

[bussdriver]

There is no clock speed. It is asynchronous and analog. Even if it had some kind of natural timing to it, some things will fire faster others slower. Chained signals will have delays along the path. The result is something without any clock speed with operations happening at the speed of analog (as fine grained as the physics allows... so in other words, crazy fast to capture it all in digital.)

Absolute precision will not be required just as analog audio doesn't need to be converted at the rate the individual molecules move and as they differ -- that level of detail is "noise" even if it is not actual random noise. You can get plenty good approximations with a decent sampling rate... but for this kind of stuff I doubt it's even 200Hz let alone 80Hz. The degree of the signal sent by neurons is not binary... so if you were thinking maybe it's 8bits... somehow I can't see how creatures which can hear better than 8bit 11 Khz audio would think at a slower rate. (ok i realize the ear is physically doing the FFT so the brain only gets the spectrum.)

Re:Analog. not digital.

[dinfinity]

Well, there's no central clock speed, but there are definitely clear instances of macroscopic synchronous behaviour in the brain:
http://en.wikipedia.org/wiki/N...

100 step rule

[NotInHere]

Its amazing:

For example a human can recognize the picture of another person in about 100 ms. Given the processing time of 1 ms for an individual neuron this implies that a certain number of neurons, but less than 100, are involved in serial; whereas the complexity of the task is evidence for a parallel processing, because a difficult recognition task can not be performed by such a small number of neurons, example taken from [zell94, p24,]. This phenomenon is known as the 100-step-rule.