nettron

Could it be that the brain doesnt actually store each individual "bit" of information ? but instead stores information as an algorithm (think recipe) which can reconstruct an "event" at will ? Remember the brain is massively parallel so it can do this quite quickly.

Ndi as pretty much summed up my thoughts on this topic but to add something to think about, as anyone ever contemplated that biological memory may not necessarily store images,sounds,smells,etc as an all-to-all bit or weight representation ? Rather it could be stored in a compressed form as an "algorithm" that is both adaptive and integrative.

No, According to your "wiring" diagram the output (axon) from neuron A is connected to the output (axon) of neuron E, according to the liturature modification occurs in the dendritic tree rather than the axons. But I think I know what you are getting at and it appears very similar to a postulate by Donald Hebb.

"...N always either an inhibitory neurotransmitter or an excitatory one no matter where you find it in the brain?" Yes.GABA and glycine are inhibitory neurotransmitters , glutamate and aspartate are excititory neurotransmitters, these will stay that way no matter where they are found in the CNS, but the effect they have at their respective receptor site can be graded.

Yes, there is a one-to-one relationship between the transmitting and receiving components of a synapse whether its an excititory or inhibitory synapse.

[/img]

Better yes but still not what was suggested , disconnect pin 2 from pin 4 and re-read other post. Also do a google search for 555 monostable .

No, the 555 (U2 ) circuit wont work as depicted in the schematic, the push-buttons ( 1 thru 6 ) should be connected to ground instead of +9Volts, check it out. Also the trigger input ( pin 2 on U2 ) likes to see a nice high to low pulse transition, pressing SWT9 will momentarily float this pin giving possibly unpredictable results , instead try connecting SWT9 (use a N.O. PB) to ground and put a resistor ( about 10K ) between pin 2 and +V. Do the same thing for the reset ( pin 4 ) but connect the resistor to ground and use a SP pussh-button instead .

...and i believe neuromorphics in VLSI to be steps in that direction.

According to the liturature this would be basically what appears to be happening between individual neurons, but tells us little about how information processing actually takes place or what coding scheme(s) the brain uses to represent, store and manipulate information. To answer a previous question, this could sum up what is stumping the scientific community at this very moment. Ok, so what could this coding scheme be ? Is it the so called rate code ? is it a timing code ? is it binary ? some other base ? is it even remotely digital at all ? Reading about how a action potential ( spike) is believed to occur, i.e. when the dendrito-synaptic inputs to a neuron is sufficient enough to overcome some membrane threshold, a stereo-typical spike is propogated down its output axon to other neurons. This does appear to be somewhat digital but there is evidense that the timing between individual spikes (called the inter-spike interval or ISI) is significant as an analog type coding scheme. If time is indeed a factor this would rule out binary in the digital sense of the word. As an analogy;recently ive bin reading about a proposed idea called rank order coding in which the firing order of an "array" of neurons can encode an enormously large input set. We are familiar with the on-off, 1's and O's binary nature of conventional computers in which information is represented by combinations of these but time has no meaning. Heres a simple illustration comparing the representation capabilities of a rank order coding scheme and binary gates. # of neurons: possible combinations: vs # of gates: possible combinations: ------------------------------------ -------------------------------- 2----------- 2 __________________________2------------- 4 3----------- 6___________________________3------------- 8 4----------- 24__________________________4------------- 16 5----------- 120_________________________5------------- 32 6----------- 720_________________________6------------- 64 7----------- 5040________________________7------------- 128 8----------- 40,320______________________8------------- 256 9-----------362,880______________________9--------------512 10----------3,628,800____________________10-------------1024 Obviously as can be seen, has the number of neurons increase the number of possible non-repeating combinations quickly increase compared to their binary counterpart. This just illustates that there are possibly other more powerful coding schemes foreign to conventional digital computing.

AFAIK, there are two primary methods: Anterograde and Retrograde, both involve some type of photographic mapping of axonal pathways rather than injecting and "listening" for a signal. A single nerve bundle can contain millions of axons, so probing to find which axon(s) are carrying the original signal would be a daunting task.

Might also want to look into the theory proposed by Karl Pribham in which he suggested that the Brain is like a holographic processor. We know that with a hologram when you cut away a section of it, we dont loose much of the original. In fact the whole original "picture" can be reconstructed from just a tiny section, which appears to make parallels with what is known in modern neuroscience, when some neurons die we dont suffer catastrophic failure...try doing that with a digital processor. Also i think it was Mead that said something to the effect " the neurons that are used for processing information are the same ones used for storing that information "....intreguing.

Might also want to look into the theory proposed by Karl Pribham in which he suggested that the Brain is like a holographic processor. We know that with a hologram when you cut away a section of it, we dont loose much of the original. In fact the whole original "picture" can be reconstructed from just a tiny section, which appears to make parallels with what is known in modern neuroscience, when some neurons die we dont suffer catastrophic failure...try doing that with a digital processor. Also i think it was Mead that said something to the effect " the neurons that are used for processing information are the same ones used for storing that information "....intreguing.

I agree that the terms "synaptic plasticity" is a rather no-brainer, its like saying plastercine (clay?) is malleable. According to the liturature, its well known and has bin proven experimentally that synapses are modifiable. Im assuming what we're getting at here is what and when. To be more specific; what biological "rules" account for synaptic modification and when do these rules come into play? Although Hebb's postulate ('49?) didnt account for depression and has bin broadened into todays LTD and LTP, seams rather sad if this is all we have. Or is there more to syncrony than one would think. To use the Cerebeller purkinje cells that i had alluded to earlier as an example; its believed that when the climbing fiber is active, only those parallel fibers that are active at the same time( or during some short time window) will cause their corresponding synapses to be facilitated. This speaks volumes for syncrony but tells us nothing about why this should occur at that particular time, after all we dont want to learn everything...only those things that are significant. The theories of David Marr and James Albus shed some light on it suggesting that the climbing fiber inputs are the training signals that "fine-tune" the motor programs from the neo-cortex from which they originate. This fine-tuning then appears to be accomblished via sensory feedback and stored motor programs( a look-up table?) so that actions can be both adjusted on-the-fly and predicted before the action is even completed. The odd thing about all this is that the output from the Cerebellum is strictly inhibitory ( puts the brakes on ) unto the deep Cerebeller nuclei. Guess this either raises eyebrows or just raises more questions

Appears what you are referring to is Hebbian learning which is really just a primitive form of learning at the single neuron level. Hebb postulated that "when an axon of cell A is near enough to excite cell B or repeatedly or consistently takes part in firing it, some growth or metabolic change takes place in one or both cells such that A's efficiency, as one of the cells firing B, is increased." Pavlov was on a similar track with his dog experiments, a form of classical conditioning involving the repeated presentation of food while ringing a bell. The dog would eventualy associate the ringing of the bell with food and would salivate. Again a very basic form of learning....believe adaption is another. I think the "real learning" you are referring to, be it cognitive or motor, would be far more involved than this. Think about the first time you tried to ride a bike or read a book. I'd say learning at this level would require a complex interplay of various neuron assemblies that each follow some ,as yet, unknown "rules" for synaptic modification. The purkinje cells of the cerebellum is a good example. http://www.robotic.dlr.de/Smagt/research/cerebellum/cerebellum.html