search

vandermeerlab / functional_connectivity

For LFP and single unit analysis of simultaneous OFC and vStr recordings on delay discounting task
2 stars 0 forks source link

Identify other freq bands (theta, beta) to lag/lead analysis in #1 #4

Open stottbot opened 4 years ago

stottbot commented 4 years ago

What frequency bands do we want to analyze for XCorr (and PSI).

stottbot commented 4 years ago

OFC-vStr coherence

stottbot commented 4 years ago

Here's the figure from before. Will do delta, beta, g50, g80. I didn't see much theta power (depended on the session), although other folks have reported theta in OFC and vStr. So maybe it makes sense to include that also. Question: I will want to choose different epoch lengths for different frequencies. I'll look at these by eye. Clearly, gamma occurs in discrete epochs, and makes sense to look at gamma EVENTS. Do we think the same applies for delta, theta, beta?

stottbot commented 4 years ago

In the Catanese paper, you guys made sure to exclude overlapping low and high gamma events. With short events (100ms), they are mostly non-overlapping anyway. Right now I am not trying to exclude overlapping events. I was planning on the same as I add other frequencies. Esp. since there is cross-amplitude coupling between delta and gamma. Does this strategy make sense? Some snippets of data will be analyzed more than once (i.e. for different frequencies).

stottbot commented 4 years ago

OFC-HIPP sessionwide coherence OFC-VSTR sessionwide coherence Here's the coherence spectra I calculated for OFC + HIPP and VSTR + HIPP. They are a bit weird, having a peak at ~12Hz. Not sure if that is a harmonic of theta, and if so, why its larger than theta.

mvdm commented 4 years ago

Question: I will want to choose different epoch lengths for different frequencies. I'll look at these by eye. Clearly, gamma occurs in discrete epochs, and makes sense to look at gamma EVENTS. Do we think the same applies for delta, theta, beta?

The "events" will be longer for delta and theta, corresponding to resting periods and runs, but I think the same approach makes sense, you don't want to estimate xcorrs if there is no power at that freq.

mvdm commented 4 years ago

In the Catanese paper, you guys made sure to exclude overlapping low and high gamma events. With short events (100ms), they are mostly non-overlapping anyway. Right now I am not trying to exclude overlapping events. I was planning on the same as I add other frequencies. Esp. since there is cross-amplitude coupling between delta and gamma. Does this strategy make sense? Some snippets of data will be analyzed more than once (i.e. for different frequencies).

I think it's a very minor issue when comparing frequencies that may overlap (in the data, and because of things like filter frequency falloff) like low- and high-gamma. Won't be a problem at all for any other freqs.

mvdm commented 4 years ago

Here's the coherence spectra I calculated for OFC + HIPP and VSTR + HIPP. They are a bit weird, having a peak at ~12Hz. Not sure if that is a harmonic of theta, and if so, why its larger than theta.

Yeah, this isn't what I would have expected either. In van der Meer & Redish 2011 we found a theta peak in HIPP + VSTR coherence. Is this averaged across all animals? Do the individual-subject plots show big differences or are they all like this?