Hello,
Using a pulse sequence with a 2 s TR, I am modeling and contrasting two "color-coded" event classes with irregular stimulus timing using the -tstim option for each subject's individual data as follows:
waver -GAM -dt 2.0 -peak 100 -tstim `cat CRPwht.1D` > whtpress2.1D
waver -GAM -dt 2.0 -peak 100 -tstim `cat CRPgrn.1D` > grnpress2.1D
This results in waveforms that correspond to the time-stamps of the events in the subject's behavioral data record. The vectors last only as many lines as the HRF of last event of each class completes, since not all subjects behave similarly. However, for all subjects, both vectors seem to run close to the 660+ TRs of the task (i.e. text editing program will show one column of 600+ rows, each with a waveform value).
However, ANFI keeps giving me the error message in 3dDeconvolve reading:
3dDeconvolve Error: Input stimulus time series file whtpress2.1D is too short.
(Since the wht event class is the first -tstim generated .1D input in the script, it may be that 3dDeconvolve might fuss about the grn too)
Any idea why this is so? The .1D time series file looks fine to me.
Jim Bjork