pvc-2 results

Has anyone managed to re-create the eigen-vectors on the space time plot for the 1d whitenoise data as illutrated in figure 2 of J. Touryan 2002?  I tried a number of the cell recordings, but to no avail.  The eigenvalues look reasonable but cannot see any patterns in the eigenvectors; I may be doing something wrong.

I attached a plot for data in folder 000601.c05.  Any comments or helps are appreciated.

I am having a difficult time reproducing the results as well. What I am getting looks like what you have. I tried just randomly sampling from the stimulus ensemble and calculating the means and principle components, the results I get look similar to the spike-triggered mean and principle components... Implying that the spikes aren't actually triggered by any specific stimuli...

I reproduced the results. After you load a few spike files from the same cell (each folder under the Spike_and_Log_files folder is from a single cell), divide the spike times by 10000 to get the units into seconds. Then look at the log for one of the spike files, it will contain a parameter called "RefreshRate", this value is the same for all the log files for the same cell but can be different for different cells. You need multiply spike times (now in seconds) by this value, this will give you what frame the spike occurred in. You will probably want round all these values UP to the nearest integer. Now you want to retrieve these frames from the stimulus ensemble. Make sure you offset your frames correctly, if you are going 16 frames into history then a spike at frame 6 won't have a complete sample point, don't accidentally use frames 6-22 in this case. Cell 000802.c05 will have a response that looks like the one in the 2002 paper, most of the other cell principal components tend to be very simple gabor type features.

Hope this helps.