Saturday, November 26, 2016

Sleep Brain Waves Are Not Only Slow, But Circular

Memories are delicate physical structures in the brain, and get cleaned up / thrown out / enhanced during sleep- but how?

The mysteries of sleep are slowly becoming unraveled, and not a minute too soon, since we are both depriving ourselves of sleep as never before, and using such deprivation as an instrument of torture. Sleep seems essential for refreshing the brain in multiple ways- tidying in both physical and logical respects. A "glymphatic" system that operates at night, when the cellular volume of the brain shrinks slightly, to physically drive cleaning fluid through the brain to clear out garbage, has been one remarkable finding. Sleep is also when memory "consolidation" happens, which means reinforcing the more important ones, re-distributing or copying them from the hippocampus to the cortex, and deleting less important ones. Then there is dreaming, that link to the deep unconscious which sleep facilitates, with perhaps higher psychological functions.

A well-known feature of sleep is its brain waves, which are especially strong and slow. They appear to function as part of the memory consolidation system, but really, we know very little about them. There are two types of significant sleep waves- the sleep spindles that happen in stage 2 sleep, and the very slow delta waves that happen in the deepest level, stage 3.

A recent paper looked at the geometric pattern of the spindle waves, using epileptic patients who have had their skulls opened extensively for investigation, and allowed the researchers to apply a large field of electrodes to one hemisphere for this tangential study. It finds that instead of the whole brain pulsating with simultaneous spikes and troughs, there is a moving, circular pattern of activation, progressing from the lower temporal cortex, to the parietal cortex, on to the frontal cortex, and back to the start. This makes sense causally, in that there are always conduction delays from one place to another, so it is difficult to imagine one pacemaker (in this case the thalamus) running simultaneous electrical oscillations all over the brain.

Example of sleep spindle propagation. A shows one spindle sequence and the electrode locations, B shows the time course, and D shows averaged vector directions of propagation over all subject and readings, notably in a temporal -> parietal -> frontal direction.

The interesting thing is that this paper argues that this pattern also makes sense functionally, in that it helps consolidate connections between distant points much better than simultaneous activation would. I find their case hard to understand as well as doubtful, but it revolves around the timing issues of long-term potentiation (LTP) and depression (LTD) among neurons.

Spike timing-dependent plasticity is the name for a broad theory of how neuronal connections are managed. If two neurons are connected to each other in the usual fashion, with dendrites from A exciting the axonal network of B, the relative timing of firing of A and B has great influence on whether their connection (synapse) is strengthened for future events, or weakened. If A fires ~10 ms (milliseconds) before B, then the synapse is strengthened. This makes obvious functional sense, implying that A *caused or helped cause B to fire, a successful event that would presumably be good to enshrine in a more permanent connection. Conversely, if A fires anywhere from 10 to 50 ms after B fires, then the logic is reversed, and the physical effect is also reversed: the synaptic connection is made weaker.

Proposed theory, whereby synchronized spindles (top) would result in troughs falling on the evoked action potentials of the original targets (EPSP), causing depression / weakening of all connections. But in B, which resembles the actual state of affairs, successive spindles, when hitting the target neurons, would be at peak value, (assuming that targets of the original neurons and the spindle wave are traveling in the same direction), and thus foster strengthening of all connections.

Given that the sleep spindle waves happen at ~11-15 Hz, or about 80 ms intervals, the authors argue that if they just propagated point to point from the thalamic pacemaker out to points in the cortex, they would arrive at various times, but their effect on secondary targets- the targets of the immediately driven cells, whose firing is delayed by, say, 20 ms- would be to cause chronic long-term depression of those target connections, since the next spindle peak falls roughly into the zone of LTD.

On the other hand, if the spindle waves propagate in a wave-front fashion through the cortex, then (B in figure) the target cells would be hit more or less simultaneously by the evoked firing from the A cells excited by the spindle wave, and then the spindle wave itself as it progressed to hit them as it moved through the cortex.

The researchers go on to find, beyond from the rotational progress of the sleep spindles, that these 11-15 Hz waves entrain gamma waves as well, and they imply that over 2.5 hours of sleep which they observed in one subject, these gamma waves strengthened in a way that supports their hypothesis that the sleep spindles are progressively reinforcing neural connections, including memories.

I find this work very hard to take seriously, though it comes from a very serious lab. If the neural network is 3-dimensional and extensive across the cortex, there is no way to predict the transmission time (estimated above at 20 ms) from one neuron to the next. Nor does the orientation of each particular sub-network necessarily have anything to do with the circular rotation scheme seen in the electrical recordings. That geometric data is much more easily explained as a mechanistic consequence, even side-effect, of not being able to activate all areas of the brain at the same time. Lastly, the idea that sleep spindles, or any process, indiscriminately strengthen all neural connections seems unhelpful, since the point of the "consolidation" process is both to discard old and minor memories / connections as well as to enhance more significant ones.

Graphs of spindle wave phase, taken 5.6 minutes apart in one subject. This suggests to the authors that coherence is progressively enhanced through the stage 2 sleep process, presumable due to positive neural connections being strengthened. Related videos and data are at the publication site.

At the same time, I do not have a counter-theory about how these waves accomplish their function, which certainly is connected with memory management. (Please comment if you have greater insight into these processes or this paper!) So we have a great deal to learn. It is a fascinating area of research, trying to build a unified theory of how the anatomical connections in the brain, and the electrical functions including oscillations of various periods, function to make our minds function, and refresh.

1 comment:

  1. My guess would be that such sleep brain waves are activating neurons that are close to firing potential - which results in activating subconscious thoughts too.

    In the same time the waves are depleting parasite potentials from neurons that don't have useful activation, preparing the brain for a new day of storing and processing new experiences. I name "parasite potentials" the neurons activation close to firing, that doesn't synchronize with others to create meaningful concepts - don't have the potential to strengthen dendrite links and don’t create chain activation.

    Maybe the sleep wave propagation path is the same with the propagation of day thoughts too.

    What do you think?

    P.S. Please excuse my English, something sounds wrong there... :)