Saturday, March 22, 2008

Oscillations Part V: Strength and efficiency in numbers

(This is a continuation of notes taken from Ginger Campbell's podcast #31 with György Buzsáki, and a continuation from Part IV):

So what is it that changes during synchronization?

* individual neurons don't change their firing rate at all
* when synchrony emerges through oscillations, it's the timing of the neurons relative to each other that changes
* the result is that now the impact of the same neurons when they are firing in synchrony is much more effective
* in fact these ripple patterns have been associated later on in several laboratories with the transferring of information from the hippocampus to the neocortex
* this was just one single oscillator

What happens when you have two oscillators?

* when two oscillators come together with slightly different frequencies, we get an interference pattern
* the best example for this interference pattern has been provided by John O'Keefe from the University College London
* he observed that individual place cells in the hippocampus oscillate at a frequency slightly faster than the ongoing so-called theta frequency oscillation
* the result of this is very precise timing - the importance of this that was translated into behaviour is that the phase position of the action potential relative to the ongoing clock cycle, reliably predicted the position of the animal
"'s a convincing case where we go from the basic features of oscillators through physiology all the way to behavioural significance. And then you can pose other questions - how would this be possible without an oscillator? And I can come up with some scenarios that could be done without oscillators, but it would be very expensive energetically and it would involve a lot more computation and would be a lot more complex."

What about when many oscillators come together?

Simple rule that applies to oscillations of various frequencies:
* the brain is a physical system
* information from one place to another goes through
1. axonal conduction delays
2. refractory delays:
"...when one neuron discharges, its target neuron will discharge later, not only because it takes time for the action potential to go through the axon, but it also, when the neuronal transmitter is released, on the post-synaptic neuron, the neural transmitter has to charge the membrane to threshold - so there is a time delay - a finite time delay of how fast the information can go from one place to another."


Brains are very slow mechanisms compared to computers:
* if it generates a fast oscillation that has a short time period, the number of neurons, or the proportion of neurons that can be involved in these fast oscillations, are relatively small
* the oscillation is typically local
* when the time period is long, there is time for activity to propagate from one area, from one set of neurons to many other sets of neurons
* a large volume is involved, the result of which is that the synchrony of the population or the proportion of the neurons that are engaged in this rhythmic activity is much larger:
"This explains the common observation that slow oscillations are always high in amplitude, or almost always high in amplitude, whereas fast oscillators are small in amplitude. But the important consequence for the purpose of our conversation is that slow oscillators can affect fast ones, in such a way that the phase of the slow oscillator, because it involves very large areas, can organize the activity of the faster ones that emerge locally."

* though other possibilities also exist to coordinate activity, oscillations provide the cheapest possible solution:
"If you talk to anybody who works in this field he'll remind you that our brain uses 20% of our energy; with a newborn it's actually 50% and if you look at the smallest mammal, the tree shrew, the tree shrew uses continuously even during sleep, 50% of all the energy to sustain brain activity. It's very expensive. Evolution must be very careful how it allocates resources."

No comments: