if they don't have any trials, SPM won't be happy if you give a list of zeros. In this case, you'll need to exclude that column from the design matrix.
I would code the dummy even exactly as you coded real events, otherwise (I've never coded as ones and zeros, I usually enter events as times for convolving with an HRF, but whatever works for your SPM method).
You can also code 'dummy' events for other things. I sometimes do this if I think something can and should be modeled to best capture the BOLD response. You don't need to contrast everything which is a part of your first level model.
Best of luck,
Colin Hawco, PhD
Neuranalysis Consulting
Neuroimaging analysis and consultation
www.neuranalysis.com
[log in to unmask]
-----Original Message-----
From: Vincent Yeo [mailto:[log in to unmask]]
Sent: January-30-18 8:35 PM
To: [log in to unmask]; Colin Hawco
Subject: Re: Question about modeling error trials
Dear Colin,
Thank you for your reply, which is really helpful. With regard to the use of dummy event, I still have a few questions:
1. I can create a dummy event of error responses for every subject, where 1 indicates error trials and 0 indicates correct trials. For subjects without error responses, the event would be an all-zero vector. Do I understand correctly?
2. Should I also use dummy variables for other events (e.g., reward anticipation in this case)?
Best,
Vincent
|