That's a different system than the cognitive parts of us. And yeah, it's fast. Very fast. We pretty much share that system with most chordates of any account.
My dog can catch food about 100 msec after it leaves my hand. Fast. but the people who know these things say that the great leap was bipedalism - once you get neural critical mass to be up on two legs, the rest is pretty easy.
Speed of critters matters more on the Discovery channel than in real life. Lotta slow ones out there.
Yes. But the computation model isn't like that. Neural systems work more like a CAM than a bespoke logic system. that's just a metaphor, it's not that much like a CAM either. It is like a CAM in that it trains.
They are *specialized cells*. Neurons are different in different contexts. So there probably is some smarts in 'em - I just don't know what that means, exactly.
Tackling just this one piece. No single cell "learns and executes strategies", that is a temporal behavior of a complete brain. Nor does any single cell "recognize grandma", a much more temporally local event. Your reduction to single cell is irrelevant, incorrect, fatuous, and completely false to facts. = It does not even rise to red herring or strawman level.
Actually a significant part of that is also seen in many sports. = Moreover there is a reason, in the 1960s i introduced to the concept of "muscle memory". After having learned rather a bit more is turns out that the brain trains the nervous matter in the spinal column to execute the complicated movements with minimal, and only general, control from the cerebral cortex. You could consider it trainable reflexes, part of what differentiates the master that can only tell to do the practice = exercises, and the one that can evolve new techniques, is this bit of understanding.
=46alse to seriously ill thought out. Synapse electrochemical = transmission times are seriously sub-millisecond. Those spacings are reasonably small (sub-micron IIRC). Spacing within connections and between connections with the general diffusion issue means that there must be local signal cancellation. Please note that nerve firing rates can reach at least a few kilopulses per second. That cannot happen if the spacing and the chemistry is not fast enough. Study up, or don't.
Nope, once again in ignorance you have it inside-out. The computation is in the network, not the cells. It is all about patterns of cells firing, probably in specific sequences (and possibly timings) to boot.
It is not the hair cells that are resonant (they are actually fairly uniform) but the structure of the cochlea itself that causes the = frequency selective localization of the excitation of the hair cells. Moreover the physical location by frequency is rather counterintuitive.
Actually there is good evidence for that, which you just don't happen to have found out about.
Irrelevant.
Some are just complex gates. Brain neurons are closer to small 22FV10s = or similar. But is more the connections with its "neighbors" and the differences between them, than the native capability that determines the total result. Again, the data is there you just haven't worked with it enough yet.
Here, yet again, you embarrass yourself. You blatantly know nothing = about neural network programs or theory. No neural network expert that i have read even tries to suppose that they can explain the brains higher functions. They are having more then enough fun and trouble thing to get them to duplicate reflexes in primitive animals. Let alone do any useful things, like control HVAC better. YMMV
Meters per second isn't much time at submircon distances.
Though the processing is in the interconnects, transmission time for fairly large "distance" (thousand of cells distance) in brain cell "locality" is comparable to CLB/cell response time. Locality is not the bugbear you would make it out to be.
Ah, they call their gadgets "neural networks" ... well ... why?
I have seen their topology, a mess of resistors and comparators, claimed to emulate the way a biological neural system works. Hence the nomenclature.
formatting link
"Artificial neural networks are composed of interconnecting artificial neurons (programming constructs that mimic the properties of biological neurons). Artificial neural networks may either be used to gain an understanding of biological neural networks, or for solving artificial intelligence problems without necessarily creating a model of a real biological system."
That the above is accepted and absurd is my point.
"A. K. Dewdney, a former Scientific American columnist, wrote in 1997, "Although neural nets do solve a few toy problems, their powers of computation are so limited that I am surprised anyone takes them seriously as a general problem-solving tool." (Dewdney, p. 82)"
Right on, AK!
--
John Larkin Highland Technology Inc
www.highlandtechnology.com jlarkin at highlandtechnology dot com
Precision electronic instrumentation
Picosecond-resolution Digital Delay and Pulse generators
Custom timing and laser controllers
Photonics and fiberoptic TTL data links
VME analog, thermocouple, LVDT, synchro, tachometer
Multichannel arbitrary waveform generators
Perhaps You should (re)read the actual legislation.
Irrelavant.=20 Amendment I
Congress shall make no law respecting an establishment of religion, or prohibiting the free exercise thereof; or abridging the freedom of = speech, or of the press; or the right of the people peaceably to assemble, and to petition the Government for a redress of grievances.
=46rom:
formatting link
Realisically it is the three days after (remorse) pill. =20 There are other drugs to use sooner (morning after pill).
Yep.
Isn't it amazing how few people have learned this?
Our disagreement is about what this Government may legally require of a religious organization (church of sponsored entity thereof) rather than what some specific religion may believe.
Yes. At least where critical paths are concerned. The blink reflex for instance relies on local processing where the optic nerves combine behind the eyes and something similar happens in the nose. I often find that I sneeze violently a second or so before I am conscious of a noxious smell like oil seed rape which is in flower at the moment.
Accuracy is far more important.
There is enough speed already. Basic neurons are good for 10m/s and the fast axions are good for >100m/s which is about a 20ms response time to touching something hot with your foot and 10ms for a hand. Most of our ability to catch balls and play tennis comes from anticipating the motion of the ball which is an altogether higher level skill.
*We* use DNA and RNA for computations in the test tube, but if life was doing it directly like that we would see vast amounts of spurious non-genome DNA floating around in cells and we don't so end of story. In fact we can use DNA fingerprinting to reliably catch criminals because the stuff is a unique marker to an individual (provided that the forensic labs do a decent job of handling it - UK has just dismantled its top research facility in this area to save money).
That isn't to say that nerve cells are simple though. There are folk modelling them with increasingly accurate computer models. The best that I could find in a quick poke around was:
formatting link
But I am not a biologist so I can't be sure how well they are really doing and what shortcomings there are.
At least computer scientists *THINK* instead of making wild unsupported assertions and futzing with the numbers like you do.
I suppose it depends here what you mean by pretty clever.
They obey fairly basic stimulus response rules which have served them well in the past. Roughly codified as eat anything you find smaller than you are and hide from anything bigger. When fat enough reproduce. Essentially a chemical finite state machine and we call it life.
[snip]
I think you will fairly quickly be proved wrong. One of the better short introductions to how current thinking in this field is running is online at:
formatting link
And hierarchical temporal memory is showing enough promise to have been deployed in several real world applications:
formatting link
It is also peppered with cranks who have an idea and would not listen to or look at any evidence that conflicts with their own prejudices.
The rough size scale is 0.1m which on the fast axons at 100m/s is 100us and on the slow channels about 1ms per hop. We know that the make a high level decision to action time from FMRI is of the order of 1s so that leaves time for somewhere between 1000 and 10000 network hops.
The neocortex is more tightly bunched and better organised than that so these numbers represent a worst case scenario and there is massive parallelism available at the front end.
And most of the connectionist community who are at present having something of a renaissance.
No about the same or at most one order of magnitude higher. Best escape reflex for amphibians and fish is of the order of 10-20ms see
formatting link
And for a human the trained reaction time to an unpredictable trigger event is ~100ms at absolute best and on the average 150-250ms.
IOW Devolving more control closer to the required action.
It makes all the difference in a musical performance since they can then spend time thinking about the subtle interpretation of the music. The additional memory stunt of top classical soloists learning whole concertos for their repertoire is at least as impressive.
There is also a matter of consigning routine things to the cerebellum where you no longer have to think about doing them. Learning to drive a car (particularly stick shift) or juggling rely on that pathway.
Locking the front door is one that is consigned to the cerebellum but can be annoying since you sometimes cannot remember doing it!
The problem with IQ as a measure of "intelligence" is that it is too broad to capture chess skill. I have known a dyslexic genius level mathematician - dyslexic in the written word but astonishing at tensor calculus so you can get exceptional individuals with innate talent in a very narrow area and huge gaps elsewhere.
I once played Nigel Short in a schools match and was soundly beaten. I did get a free coaching lesson in the Sicilian out of it though. The difference between us was that he didn't need to look at the scoresheet at all to rearrange the pieces to any previous position from the game.
Chess is a game for innate pattern matchers (and Go even more so).
As a reasonably strong chess player I would have to say here that correlation does not imply causation. The most obvious cause of the trend described is that only the very strongest players can afford to make a living at it and so they play many more games than the rest.
It is also true at the bottom end that players who can't make progress get bored and give up. BBC's More or Less had an interesting program snippet about the shortage of top women chess players (only one in the top world 100 - Judith Polgar) last week. Basically too few girls start playing chess and they don't find the competitive game attractive whereas boys do. To be world class at chess you have to start early.
formatting link
From 15:55 onwards. You might find the previous bit about the Greek railways amusing too.
I reckon with modern computer training methods you could get anyone who isn't completely thick up to a respectable 1600 ELO chess rating.
Fairy nuff. I remembered it wrong by an order of magnitude.
Electrons have three fundamental properties, mass, charge and magnetic moment. The magnetic moment is just like a little magnet. In a magnetic field the lowest energy state is when the MM is aligned with the field. In quantum mechanics these staes become qunatized.
Correct. The hair cells sit on a basilar membrane that is (counter-intuitively) narrow at the base end of the cochlea (which is conspicuously larger on the outside) and wider at the top (which appears tiny from outside). The trick is that the active part of the membrane is much smaller than the apparent outside width of the cochlea, supported by bony ridges that have a larger gap between them as you go up.
The basilar membrane has separate fluid spaces on each side. (It's actually a bit more complicated, but that's the crux of it.) The vibrating eardrum activates a 3-bone chain that ends with a plunger in the "oval window" that acts on the fluid on one side of the basilar membrane, while pressure is relieved at the "round window" membrane on the other side of the basilar membrane.
This sets up vibrations in the basilar membrane, and due to the reverse taper (high stiffness and low weight at the start, going to low stiffness and higher weight as you go up) the result is a traveling wave whose shape is related to the stimulus frequency. At high frequencies, the wave peaks near the origin, while low frequencies peak farther along. (There used to be furious debates about whether this was really "resonant" behavior, versus traveling wave behavior,or both.)
The peak is astoundingly sharp... equivalent electronic Q values in the 100s to 1000s or more. (In auditory work they use a Q based on -10 dB instead of -3 dB, so the numbers you see in papers look smaller.) People used to argue about how this could possibly be true, given knowledge of the mechanical properties of the components. But it is now known to be due to an active feedback process.
Along the basilar membrane are 3 rows of outer hair cells and one row of inner hair cells. We knew for a long time that essentially only the inners were the ones the brain listened to. The outers only got signals *from* the brain, which served as a gain control.
But the sharpness of the tuning was due to electromotility of the outers, sort of like piezoelectric elements. Vibration caused voltage which caused more vibration, and away it went. This only happens at very low levels, near threshold, where the measured Q is so high. At higher levels things the electromotility saturates and you get "normal" linear system behavior.
Amazing system!
Bob Masta DAQARTA v6.02 Data AcQuisition And Real-Time Analysis
formatting link
Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, FREE Signal Generator Pitch Track, Pitch-to-MIDI Science with your sound card!
The original was quickly modified after the bishops complained. The current version is as I have stated.
This is not about what a religion may require of its adherents, but about what a public employer may require of its employees, who may or may not not share the same religion as the employer. Your boss can't force you to practice his faith. The bishops want to force their employees (Muslim, Protestant, whatever) to adhere to the bishops' religious tenets. That is what this is all about.
Bob Masta DAQARTA v6.02 Data AcQuisition And Real-Time Analysis
formatting link
Scope, Spectrum, Spectrogram, Sound Level Meter Frequency Counter, FREE Signal Generator Pitch Track, Pitch-to-MIDI Science with your sound card!
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.