Whitney,

So you set up 3dDeconvolve with 2 regressors: real pictures and nonsense pictures?

> 1) How can I contrast naming pictures, whether
> real or nonsense, with rest? Do I simply need to
> create a stimulus timing file for rest "trials" to input
> to 3dDeconvolve?

No, what you want is an F test in -glt with a matrix of 2 rows: one row with a 1 for real pictures (and 0's for the rest), and the other with a 1 for nonsense pictures (and 0's for the rest).

Or if your stimulus files are fed in 3dDeconvolve as

-stim_label 1 RealPic
-stim_label 2 NSPic

you can simply set up the desirable test as

-gltsym 'SYM: RealPic \ NSPic ' -glt_label Pic

> Is it possible in AFNI to put one of the conditions
> aside and contrast the other condition with rest?
> For example, can I compare naming real pictures
> with rest, without taking into consideration
> activation for nonsense pictures?

3dDeconvolve should automatically generate such a contrast as part of the output for you without any further effort.

Gang