Hi Rick,
Thanks for replying me!
I thought the longer ISIs the better the deconvolved results. And I thought we needed a slow event design to get better signal change curve of each stimulus. So the long rests were expected. But I am not sure about this. I thought the distribution of events should look evener but some ISIs are too small and the others are too big.
I also tried to reduce run_lengths to 180 includes pre_rest and post_rest:
set num_stim = 3
set num_runs = 4
set pre_rest = 0 #pre_rest will be added manually later in the formal experiment
set post_rest = 10
set min_rest = 10
set tr = 2.0
set stim_durs = 2
set stim_reps = 8 # This specifies the number of repetitions of each stimulus type, per run
@ run_lengths = 180 * `printf '%.f' "$tr"`
The optimal sequence looks evener. But the post_rest is still too long to take for me:
timing_tool.py -multi_timing stimes.${iter}_0* \
-run_len $run_lengths -multi_stim_dur $stim_durs \
-multi_show_isi_stats
ISI statistics (3 elements) :
total per run
------ ------------------------------
total time 1440.0 360.0 360.0 360.0 360.0
total time: stim 192.0 48.0 48.0 48.0 48.0
total time: rest 1248.0 312.0 312.0 312.0 312.0
rest: total isi 1142.0 290.0 280.0 280.0 292.0
rest: pre stim 18.0 0.0 12.0 6.0 0.0
rest: post stim 88.0 22.0 20.0 26.0 20.0
num stimuli 96 24 24 24 24
min mean max stdev
------- ------- ------- -------
rest: pre-stim 0.000 4.500 12.000 5.745
rest: post-stim 20.000 22.000 26.000 2.828
rest: run #0 ISI 10.000 12.609 36.000 5.606
rest: run #1 ISI 10.000 12.174 20.000 2.480
rest: run #2 ISI 10.000 12.174 16.000 2.329
rest: run #3 ISI 10.000 12.696 22.000 3.548
all runs: ISI 10.000 12.413 36.000 3.674
And the NSD of optimal sequence becomes bigger (from 0.332600 to 0.354300).
Do you have suggestions for me to improve?
Yu Zhang
Edited 4 time(s). Last edit at 05/07/2018 05:32AM by Zhang Yu.
