Thanks. I don't think I used -iresp before - I made a mask for each ROI, took the contrast .BRIK (e.g. conditionVbase, taken from 3dDeconvolve with 3dmerge) and used 3dmaskave to get the average timeseries in that ROI for that condition. I'm not sure if that was correct? In any case, here is my 3dDeconvolve - thanks again:

3dDeconvolve \
-input scaled_allruns+orig \
-mask mask+orig \
-concat allruns.1D \
-censor 'FA09_idealfile.txt[5]' \
-num_stimts 5 \
-stim_file 1 'FA09_idealfile.txt[0]' -stim_label 1 posneg \
-stim_file 2 'FA09_idealfile.txt[1]' -stim_label 2 posneu \
-stim_file 3 'FA09_idealfile.txt[2]' -stim_label 3 negneu \
-stim_file 4 'FA09_idealfile.txt[3]' -stim_label 4 neuneu \
-stim_file 5 'FA09_idealfile.txt[4]' -stim_label 5 visvis \
-stim_minlag 1 0 -stim_maxlag 1 6 \
-stim_minlag 2 0 -stim_maxlag 2 6 \
-stim_minlag 3 0 -stim_maxlag 3 6 \
-stim_minlag 4 0 -stim_maxlag 4 6 \
-stim_minlag 5 0 -stim_maxlag 5 6 \
-num_glt 15 \
-glt 1 posnegVbase.glt -glt_label 1 posnegVbase \
-glt 1 posneuVbase.glt -glt_label 2 posneuVbase \
-glt 1 negneuVbase.glt -glt_label 3 negneuVbase \
-glt 1 neuneuVbase.glt -glt_label 4 neuneuVbase \
-glt 1 visvisVbase.glt -glt_label 5 visvisVbase \
-glt 1 posnegVvisvis.glt -glt_label 6 posnegVvisvis \
-glt 1 posneuVvisvis.glt -glt_label 7 posneuVvisvis \
-glt 1 negneuVvisvis.glt -glt_label 8 negneuVvisvis \
-glt 1 neuneuVvisvis.glt -glt_label 9 neuneuVvisvis \
-glt 1 posnegVneuneu.glt -glt_label 10 posnegVneuneu \
-glt 1 posneuVneuneu.glt -glt_label 11 posneuVneuneu \
-glt 1 negneuVneuneu.glt -glt_label 12 negneuVneuneu \
-glt 1 posnegVposneu.glt -glt_label 13 posnegVposneu \
-glt 1 posnegVnegneu.glt -glt_label 14 posnegVnegneu \
-glt 1 posneuVnegneu.glt -glt_label 15 posneuVnegneu \
-iresp 1 irf_posneg.irf \
-iresp 2 irf_posneu.irf \
-iresp 3 irf_negneu.irf \
-iresp 4 irf_neuneu.irf \
-iresp 5 irf_visvis.irf \
-fitts decon_fivecond_fit.fit \
-fout -tout -rout -bucket decon_fivecond