InstaCorr and FIM

[ns1]

Hi AFNI users,
I’m trying to use - just as exploratory analysis - InstaCorr on the waveforms from BLOCK in 3dDeconvolve…

When I run the InstaCorr on the visual cortex (image 1), the output correctly shows me positive and negative correlations.

When I do the same thing from the thalamus (image 2), negative correlations are shows as positive.

Am I missing something or doing something wrong?

thanks
ns

[rickr]

Hi NS,

Looking at the graphs, it is not clear to me what is supposed
to be negative that is showing up positive. For example, both
graphs show a positive response, so they should show strong
positive correlations.

But consider, these correlations should all be 1 or -1, since
they seem to be based on iresp of a fixed shape BLOCK
function. The plots all look exactly the same, and the
differences you are seeing are simply truncation errors from
something.

Along those lines, is that cubic detrending in there? The iresp
time series are short, making them more susceptible to
artifacts from something like detrending and computational
truncations. Is one cubic and the other quadratic? Is is hard
to see…

• rick

[ns1]

Sorry Rich - The attached graphs didn’t show what I meant to show.
Attached find 1 image where I have selected one voxel in the visual cortex which had a positive BOLD change (img1.png). Regions with high positive correlations
(red/orange) are depicted in the left side of img1.png and show a positive BOLD waveform. BUT, in the thalamus, regions with positive correlation (red/orange) - the BOLD waveform shows a decreasing BOLD waveform (img2.png, right) , which seems as if it should show a negative (blue) correlation?
In the visual cortex, it works correctly: If I pick a red/orange region with high positive correlation
it shows a positive waveform (img2.png, left) and if I pick a region with blue blob, it shows a negative waveform (img2.png, right)

In all cases, I am looking at the correlation of the same, positive-going voxel with these regions.

So, my question is: Why is there a red/orange blob suggesting a positive correlation when the actual
waveform in that region shows a decreasing BOLD response when correlated with a voxel showing
an increasing BOLD response.