Thanks, Daniel! I found the problem. The scanning plans for the two scans were not identical. This could be the mistake by the scanning technician.
BTW, I think I found a bug in 3dinfo.
When I run "3dinfo -extent FILE1" and "3dinfo -extent FILE2", it shows the extent information correctly; when I run "3dinfo -extent FILE1 FILE2", it shows the extent information for the two files with the extents from FILE1.
My AFNI version is 18.0.11.
Below is the terminal information.
>> 3dinfo -extent -d3 ${rootpath}/data_proc/${sbj}/s1/s1.nii
-107.147003 109.415497 -123.905998 92.656502 -41.609699 83.190308 3.437500 -3.437500 3.900000
>> 3dinfo -extent -d3 ${rootpath}/data_proc/${sbj}/s3/s3.nii
-107.694000 108.868500 -122.811996 93.750504 -39.969002 84.831001 3.437500 -3.437500 3.900000
>> 3dinfo -extent -d3 ${rootpath}/data_proc/${sbj}/s1/s1.nii ${rootpath}/data_proc/${sbj}/s3/s3.nii
-107.147003 109.415497 -123.905998 92.656502 -41.609699 83.190308 3.437500 -3.437500 3.900000
-107.147003 109.415497 -123.905998 92.656502 -41.609699 83.190308 3.437500 -3.437500 3.900000
>> 3dinfo -extent -d3 ${rootpath}/data_proc/${sbj}/s3/s3.nii ${rootpath}/data_proc/${sbj}/s1/s1.nii
-107.694000 108.868500 -122.811996 93.750504 -39.969002 84.831001 3.437500 -3.437500 3.900000
-107.694000 108.868500 -122.811996 93.750504 -39.969002 84.831001 3.437500 -3.437500 3.900000
>> afni -version
Precompiled binary macosx_10.7_local: Jan 27 2018 (Version AFNI_18.0.11)
-- Nan