Linear trend across blocks of trials at subject level

AFNI Message Board
1

Hello afni experts,

I’d look to see if there’s a linear trend across 8 blocks of trials at the subject level.
I would like to have the slope across the 8 blocks per subject and then run t-test on the slopes.
I’m not sure how to code 3dDeconvolved for that.

1- Use 8 -stim_times_AM2 option (1 stimfile for each of the 8 blocks) with the :1, :2, …:8 added after the timing for each block. But I don;t know how to look for the slopes with that option as I will still have 8 (well 16 in fact) regressors.
stimfile example:
Block_1.1D
0:1
5:1

Block_2.1D
100:2
108:2

Block_8.1D
2500:8

2- Use 1 -stim_times_AM2, but with ALL the events in this single file with the :1, :2, …:8 added after the timing for each blocks.
stimfile example:
0:1
5:1
100:2
108:2
…
2500:8

Which option or others should I go for?

Thanks for your time and advice,

pat

2

Hi Pat,

Case 2 seems appropriate to me, with all modulators and timing
in a single file. It would result in 2 betas, one for the average effect
and one for the slope (the incremental effect of the modulator).

  • rick
3

Hi Rick,

That seem to have worked.
But, I now realize that I will not be able to look at or plot in a figure the betas for each block, as I only have the slopes; but this is what I asked for in the first place so we’re good here :).

But, I (maybe my boss or a reviewer) would love to see these beta for each block.
So can I get these by asking for 3dDeconvolve to estimate each block as I would do with the stim_times option at the same time as estimating the slope with stim_times_AM2; so I’d have 8 stim_times (1 per block) and 1 stim_times_AM2 (for the slope) in the same script? Or do I need to run another 3dDeconvolve estimating each block?

thanks

4

Hi Pat,

Before forgetting again, note that amplitude modulators
should be listed with '’ rather than ‘:’, which is for
duration modulators, i.e. 1002 .

To list all of the betas, I would like to suggest using
-stim_tims_IM (individual modulation), which would give a
beta per event. However there seem to be 2 events per
block, which means _IM does not quite work (it would give
16 betas instead of 8).

So in order to get 8 betas, one per block, you should
break them into 8 separate stim timing files (with no
modulators).

  • rick