Monday, May 15, 2006

Explodatory sharp-wave ripples   posted by Coffee Mug @ 5/15/2006 11:05:00 PM

Yo.. Ya’ll can’t do it like ‘Tay do it.. Can’t stop, won’t stop like a Ford Explorer without the brake fluid.. - Phonte

Around the same time as Foster and Wilson were reporting that recent event sequences in the hippocampus replay in reverse during subsequent sharp-wave ripple (SWR) events, O’Neill and colleagues were reporting that SWRs occur even as a rat is wandering around and that these seem to lead to more coordinated firing in a post-exploration sleep session. I didn’t quite explain what SWRs were before. The “ripple” part is the 140-200 Hz oscillation measured in the CA1 sub-region of the hippocampus. The sharp-wave (a self-organizing burst of activity) begins in the recurrent networks of the CA3 region from which neurons project to the CA1. The strong stimulation from the sharp-wave induces synchronous, oscillatory CA1 and CA3 firing.

SWRs can happen during exploration, usually in brief pauses in theta oscillation (the predominant network activity during exploration ~10 Hz oscillation) when the rat is slowing its pace, but can also happen overlaid with theta. Rats are enjoying a leisurely stroll across campus, lost in thought. These exploratory ripples (eSWRs) are smaller amplitude, shorter, and less frequent, but with a higher oscillation frequency than ripples during immobility (iSWRs) and sleep (sSWRs).

CA1 pyramidal cells are brought into tighter synchrony during SWRs than during theta oscillations, or at least more of them fire together in short time windows. Synchrony seems to indicate some sort of network level organization and communication between synched cells. They also determine that place rate-map correlation (the tendency for a place cell to fire more in one spot over others) is worse during eSWRs compared to during theta oscillation. I think the point of all this is to show that eSWRs can muck up the current place field map (potentially to change it?).

That is not to say that eSWRs don’t have place-specific firing properties. In fact, the authors go to great pains to show that eSWRs increase firing rates in a place-specific fashion. The peak firing rate for a place cell inside its place field during an eSWR is actually higher than that from adding (“in field” baseline firing rate) + (eSWR peak rate out of field). I take this as saying that place cells that are “in field” have some special response to the eSWR, which perhaps identifies them for stronger integration into the current assembly.

The meat of the paper is in the second to last figure (figure 7, if you have access to Neuron), which shows a ton of comparisons of correlation coefficients, which makes it very confusing to me. They are looking at shared properties of cell-pairs as a sort of proxy for network cohesiveness. Cell-pairs are grouped by correlation on one measure and then tested for correlation on another similar measure for the most part. One thing that bothers me is why on earth they would take a continuous variable like a correlation coefficient and set a seemingly arbitrary threshold for dividing cells-pairs into groups rather than just analyzing it by regression.

The categorization variable is the instantaneous firing rate correlation (IFRC). This is exactly what it sounds like: How similar is the firing rate at various times between these two cells? When cells are grouped into hi-IFRC or low-IFRC groups during eSWRs and then tested for IFRC during sSWRs before (referred to as pre-sleep) or after (post-sleep) the exploration session, the first thing you notice is that IFRCs are generally weaker during sSWRs (less coherent network activity?). The second thing you notice is that the difference between hi- and lo- IFRC pairs is much bigger after exploration when the rat is sleeping in the explored environment than before the rat was ever exposed to this environment. The same thing happens if you group cell pairs by place-field similarity rather than IFRC. It seems that something is happening during exploration to arrange cells into place representing assemblies that are then reactivated in subsequent sleep.

The story here seems to be that when a rat first enters a context it doesn’t really have a map for it set up. As he wanders around a little, place cells start to sort of arbitrarily pick spots. Rather than having the place cells just sort of fluctuate in and out of representing this spot or another, these SWRs come in and whack the system so hard that they alter synaptic weights. Cells that were inching towards representing the same place have an opportunity to link-up and coordinate firing a little, builiding assemblies, whereas cells that were borderline get the boot. Previously, this was presumed to happen in sleep and was supposed to coordinate hippocampal cells with neo-cortical cells in order to consolidate the spatial memory, but this paper and the previous one I covered (Foster and Wilson, 2006) show that cells can do this during waking hours. This is probably good if you went somewhere this morning and want to return in the evening, but it deprives sleep of a special role, so sorry folks we still don’t know what sleep is for.


O’Neill J. Senior T. Csicsvari J. 2006. Place-selective firing of CA1 pyramidal cells during sharp wave/ripple network patterns in exploratory behavior. Neuron. 49:143-155.

Foster DJ. Wilson MA. 2006. Reverse replay of behavioural sequences in hippocampal place cells during the awake state. Nature 440:680-683.

PDFs for you:
Here are PDFs of the O’Neill et al. and Foster and Wilson papers.