Adding a parametric regressor to the baseline

Howdy,

I am attempting to add two parametric regressors to baseline in the deconvolution step of my analysis. I’m not sure how to do it. Any ideas?

BACKGROUND: We’re having participants read text and are using behavioral responses in the form of eye movements to build our response function. As part of that there are some basic linguistic features I would like to include in the baseline function as they are of no interest but must be accommodated for. My 3dDeconvolve command is below. I thought I could add the parametric regressors to baseline using “-stim_times_AM2” followed by “-stim_base” (see the 7th stim time series) but this errors out as illegal. I formatted the timing file as follows *.

Thoughts?
-Thanks!

=========================

3dDeconvolve
-input $subj_DIR/afni_data/epi1_volreg+orig $subj_DIR/afni_data/epi2_volreg+orig $subj_DIR/afni_data/epi3_volreg+orig
-mask $subj_DIR/afni_data/struct_mask+orig
-polort A
-num_stimts 11
-stim_file 1 “$subj_DIR/motion/motion.txt[0]” -stim_label 1 “Roll” -stim_base 1
-stim_file 2 “$subj_DIR/motion/motion.txt[1]” -stim_label 2 “Pitch” -stim_base 2
-stim_file 3 “$subj_DIR/motion/motion.txt[2]” -stim_label 3 “Yaw” -stim_base 3
-stim_file 4 “$subj_DIR/motion/motion.txt[3]” -stim_label 4 “dS” -stim_base 4
-stim_file 5 “$subj_DIR/motion/motion.txt[4]” -stim_label 5 “dL” -stim_base 5
-stim_file 6 “$subj_DIR/motion/motion.txt[5]” -stim_label 6 “dP” -stim_base 6
-stim_times_AM2 7 ${TIMING_FREQ} ‘dmBLOCK’ -stim_label 7 “freq” -stim_base 7
-stim_times_AM2 8 ${TIMING_LENG} ‘dmBLOCK’ -stim_label 8 “leng” -stim_base 8
-stim_times_AM2 9 ${TIMING_POS} ‘dmBLOCK’ -stim_label 9 “POS”
-stim_times_AM2 10 ${TIMING_LSA} ‘dmBLOCK’ -stim_label 10 “LSA”
-stim_times_AM2 11 ${TIMING_ORTHO} ‘dmBLOCK’ -stim_label 11 “ORTHO”
-num_glt 3
-gltsym ‘SYM: POS’
-glt_label 1 POS
-gltsym ‘SYM: LSA’
-glt_label 2 LSA
-gltsym ‘SYM: ORTHO’
-glt_label 3 ORTHO
-censor “$subj_DIR/motion/motion_censor_vector.txt[0]”
-nocout -tout
-bucket predictability_deconv
-xjpeg predictability_design.jpg
-jobs 2
-GOFORIT 12

From modeling perspective, it does not matter whether you consider a response as an effect of interest or no interest. So, just change the following two lines

-stim_times_AM2 7 ${TIMING_FREQ} ‘dmBLOCK’ -stim_label 7 “freq” -stim_base 7 \
-stim_times_AM2 8 ${TIMING_LENG} ‘dmBLOCK’ -stim_label 8 “leng” -stim_base 8 \

to

-stim_times_AM2 7 ${TIMING_FREQ} ‘dmBLOCK’ -stim_label 7 “freq” \
-stim_times_AM2 8 ${TIMING_LENG} ‘dmBLOCK’ -stim_label 8 “leng” \

Also, to avoid potential complexity, it might be better to use dmUBLOCK instead of dmBLOCK.

Awesome! Thank you for the quick response!

-Ben