Video link if not working here: http://archive.org/details/RobertSapolsky-BiologyAndHumanBehavior
Recurrent collateral - neuron sends an inhibitory signal back to itself. This sharpens the signal over time by concluding the neuron's message. Lateral inhibition - inhibiting neurons around it to sharpen the image by silencing extraneous info. This provides clarity as to who's speaking and when they're done.
Heubel and Weisel - dots light stimulating neurons in the optic system. So one photoreceptor in the retina and one neuron gets excited. Layer two neurons seemed to know about straight lines. Then layer 3 seemed to know about moving straight lines and curves. The thinking then being that the upper levels of complexity would know one thing, for example the way your grandmother looks at a certain angle. These are known as grandmother neurons.
To this day, evidence is lacking for establishing that the neurons work this way. The math makes it clear that this is impossible since the layers would soon need way too many neurons to fit in the brain. Instead the current thinking is that it may be about neural networks. Information is not contained in a single neuron or synapse; it's contained in patterns of excitation.
One neuron might know Monet, another Picasso, a third about Impressionism. These are connected to each other. This is the tip of the tongue stuff. Evidence that networks are the key is gathered from early stage Alzheimer's patients; the info is still in there it's just harder to get to. The worse the damage is, the more priming that has to be done to pull it forth. As the neurons die, it takes more effort to pull the info.
The lateral inhibition pathway works in pain systems. A quick pinprick encourages a fast wave, then it's stopped. If it isn't stopped you get the throbbing, chronic pain.
A sudden sharp pain turns off throbbing pain for a while.
Heubel and Weisel - dots light stimulating neurons in the optic system. So one photoreceptor in the retina and one neuron gets excited. Layer two neurons seemed to know about straight lines. Then layer 3 seemed to know about moving straight lines and curves. The thinking then being that the upper levels of complexity would know one thing, for example the way your grandmother looks at a certain angle. These are known as grandmother neurons.
To this day, evidence is lacking for establishing that the neurons work this way. The math makes it clear that this is impossible since the layers would soon need way too many neurons to fit in the brain. Instead the current thinking is that it may be about neural networks. Information is not contained in a single neuron or synapse; it's contained in patterns of excitation.
One neuron might know Monet, another Picasso, a third about Impressionism. These are connected to each other. This is the tip of the tongue stuff. Evidence that networks are the key is gathered from early stage Alzheimer's patients; the info is still in there it's just harder to get to. The worse the damage is, the more priming that has to be done to pull it forth. As the neurons die, it takes more effort to pull the info.
The lateral inhibition pathway works in pain systems. A quick pinprick encourages a fast wave, then it's stopped. If it isn't stopped you get the throbbing, chronic pain.
A sudden sharp pain turns off throbbing pain for a while.
