SoylentNews is people
SoylentNews
The brain has more computational capacity than previously thought, according to UCLA researchers:
Dendrites have been considered simple passive conduits of signals. But by working with animals that were moving around freely, the UCLA team showed that dendrites are in fact electrically active — generating nearly 10 times more spikes than the soma (neuron cell body). The finding, reported [DOI: 10.1126/science.aaj1497] [DX] in the March 9 issue of the journal Science, challenges the long-held belief that spikes in the soma are the primary way in which perception, learning and memory formation occur.
"Dendrites make up more than 90 percent of neural tissue," said UCLA neurophysicist Mayank Mehta, the study's senior author. "Knowing they are much more active than the soma fundamentally changes the nature of our understanding of how the brain computes information. This is a major departure from what neuroscientists have believed for about 60 years," said Mehta, a UCLA professor of physics and astronomy, of neurology and of neurobiology.
Because the dendrites are nearly 100 times larger in volume than the neuronal centers, Mehta said, the large number of dendritic spikes taking place could mean that the brain has more than 100 times the computational capacity than was previously thought.
Is that your final answer?
(Score: 1) by Ethanol-fueled on Tuesday March 14 2017, @11:44PM (5 children)
It seems to be a matter of elegant simplicity that the connection branches rather than the "brain" of the cell would perform the processing more analogous to a series of reflexes rather than all the action happening within the cell body. But I think that "computational capacity" is meaningless unless you're talking something like Johnny Mnemonic or ripping the brains out of the body and harvesting them as computers.
Sure, you can have smart motherfuckers...but are they conscious of everything that goes on in their heads, or can they treat it as a "memory palace" to recall any experience? Hell no, because to be granted those abilities would likely be too overwhelming to function and you'd be left with Minority-Report style super-autists living in milk baths or whatever it is they do all day.
(Score: 3, Interesting) by Runaway1956 on Wednesday March 15 2017, @02:43AM (4 children)
"treat it as a "memory palace" to recall any experience?"
Memory is funny. A lot of minds don't seem to work like everyone else's minds.
Suppose that you attended a rally. You personally get to speak to the famous person for whom the rally is being held, for less than a minute. You do realize this famous person meets hundreds of people per day. So, you got to speak to a famous person, for some seconds.
Then, sevreal years later, you again meet this famous person. He addresses you by name, and asks how things are going in your home town.
Is that remarkable? Isn't that a sign of something akin to an eidetic memory? Some people don't forget stuff.
What would really be great, is if science begins to understand how those different minds are different - and help the rest of us to remember things just as well.
Oh - the famous person in this scenario is Bill Clinton, and the not-famous person is one of my co-workers. Believe his story or not, but I've never found a reason to distrust him. I believe his story about meeting Clinton twice.
Abortion is the number one killed of children in the United States.
(Score: 2, Funny) by Anonymous Coward on Wednesday March 15 2017, @09:25AM
That is because Bill Clinton is a reptile. They have many times the memory capacity of a human being.
(Score: 2) by tibman on Wednesday March 15 2017, @01:59PM (2 children)
Bumping into a president long enough to have a conversation would be a fantastic oddity. Bumping into the president a second time years later and the president knows you by name and asks "how's your home town" is nightmare fuel. You are living on the outer limits of the twilight zone. http://shieldheadwear.com/ [shieldheadwear.com]
SN won't survive on lurkers alone. Write comments.
(Score: 2) by Runaway1956 on Wednesday March 15 2017, @03:18PM (1 child)
Well, Mike didn't "bump into" then-Governor Clinton. Mike attended a rally, where a person might well expect to "bump into" the person for whom the rally was organized.
The second time, if I recall Mike's story correctly, was during a Clinto re-election affair. But, yeah, it's still kinda scary that the then president actually remembered the name and hometown of some nobody. To be fair, Mike's hometown is only - uhhhh - let me check - 38 miles from Hope. It's possible that Clinton remembered Mike due to some shared knowledge that Mike has forgotten about, or neglected to tell me.
Abortion is the number one killed of children in the United States.
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @04:12PM
That seems more likely and a lot less scary. Unlike the other AC up there, I remain convinced that he isn't a lizard person. His "wife" is, but he just has too many mammalian proclivities to be one.
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @12:05AM (4 children)
Seems like the singularity just got pushed off awhile...
(Score: 2) by takyon on Wednesday March 15 2017, @12:40AM (2 children)
Funny you should say that, check out the comments from the source:
And Kurzweil has apparently brought the date up to 2029 [nextbigfuture.com], from ~2045.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 1) by moondoctor on Wednesday March 15 2017, @02:12AM (1 child)
>1014 vs. 1016 calculations per second
Ridiculous. The brain don't do calculations per second it works in a fundamentally different way. It's all speculation, there is not a single person that can tell you how the brain actually figures out that 2 plus 2 is 5.
(Score: 3, Interesting) by TheLink on Wednesday March 15 2017, @03:31AM
there is not a single person that can tell you how the brain actually figures out that 2 plus 2 is 5.
And that shows we aren't that close to building stuff that replicates what the brain does.
The AIs that people have built so far make rather different mistakes from the sort humans make. Look at the mistakes Watson makes. Look at the mistakes image recognition algos make: https://www.wired.com/2015/01/simple-pictures-state-art-ai-still-cant-recognize/ [wired.com]
The mistakes they make prove the AIs don't actually understand stuff yet. Heck a crow with its walnut sized brain probably truly understands more about the world than all the AIs combined: https://www.youtube.com/watch?v=cbSu2PXOTOc [youtube.com]
And to me that indicates the scientists and engineers trying to build "replicas" don't really understand how brains do what they do either. They are at the "alchemy" stage - they know how to get useful results, but they don't really have a theory that's close to reality.
(Score: 1) by moondoctor on Wednesday March 15 2017, @02:03AM
Apples and oranges. As you point out, trying to replicate the human brain is silly with our current understanding. Quantum machine learning writing it's own code (or some other approach) however...
(not dismissing the brain modelling work out of hand, there is great stuff being done these days)
(Score: 2) by martyb on Wednesday March 15 2017, @12:24AM (3 children)
The "smarts" are in the wiring? This turns everything I've learned about how the brain works on its head (bad puns intended!)
This is like saying that all the chips on a motherboard only do 10% of the work — the real processing is done by the traces!??!
Just imagine what must be going through the minds of real neurosurgeons.
Wit is intellect, dancing.
(Score: 2) by Runaway1956 on Wednesday March 15 2017, @02:48AM (1 child)
The analogy doesn't really work because the traces on the motherboard aren't 90% of the components in/on the mainboard. I never gave a lot of thought to the metal distribution inside the computer, but off-hand, I'd say the traces are probably 40% or less. (Discounting the case, the mounts, and housings for hard drives - we're only concerning ourselves with the mainboard and it's components.)
Abortion is the number one killed of children in the United States.
(Score: 2) by martyb on Wednesday March 15 2017, @07:38PM
Yeah, not the best analogy. Forgive an old graybeard who remembers learning programming on a PDP-8 minicomputer the size of a refrigerator. Its logic was constructed of Flip Chips" [wikipedia.org]. It also had a whopping 24K 12-bit words of core memory [wikipedia.org]. Then came "personal" computers like the IMSAI-8080. Back in those days, the traces on the motherboard were much larger, and the discrete(!) components (like resistors, caps, and TTL chips) were far less 'capable'. Things are a tad bit different these days, as you kindly pointed out.
So, I stand/sit corrected, and thank you for a trip down memory lane, as well!
Wit is intellect, dancing.
(Score: 2) by linkdude64 on Wednesday March 15 2017, @05:53PM
"Just imagine what must be going through the minds of real neurosurgeons."
While in surgery? Hopefully not how shakingly angry they are that they lost a bet over sportsball.
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @01:03AM
Who knew that thought was so different from general brain activity on a cellular level?
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @02:11PM (1 child)
I was taught that the various dendrites are stimulated, the signals travel to the soma where they are summed up, and then if a threshold is crossed a signal is sent down the axon to stimulate the dendrites of the downstream neurons. The observations described in the summary are in no way in conflict with that.
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @05:46PM
They seem to be saying that dendrites can cause an action potential without creating activity in the soma. I can't read the paper, so I'm not sure what kind of evidence they have for that, or whether it's just LTP.
(Score: 0) by Anonymous Coward on Wednesday March 15 2017, @06:06PM
This kind of article is usually pretty bad. All of the commentary about what this means is just conjecture meant to catch your eye. This isn't going to change artificial brain design because nobody is designing brains at this level of detail. State-of-the-art neural nets don't have stateful neurons, so the only relevance this has to them is in the synaptic weights, which they already cover. Same goes for the computational cost: the cost has always been at the synapse, which is where the dendrites get their inputs.
The analog/digital commentary doesn't really mean anything either.
What they are saying is the soma was thought to be responsible for determining an action potential (neuron firing) after being stimulated by inputs from the dendrites, but this research suggests that a dendrite could alone cause the action potential without involving the soma. Without reading the paper I can only guess at how that's meant to happen (or in what sense it's true) since the soma is effectively between the dendrites and the axon. On the other hand, a lot of stuff goes on in the soma during an AP, and bypassing that activity could have a significant impact on its state.