 |
 |
Hi Andrew, You might find this paper useful when deciding which method to use: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2654219/ Best wishes Steve -- Stephen M. Fleming PhD Wellcome Trust Postdoctoral Fellow Centre for Neural Science, NYU http://web.me.com/stephen_fleming/web/Welcome.html On 30 Jan 2012, at 06:31, Andreas Finkelmeyer wrote: > Hello Andrew, > > I’m no expert, but my guess is that using a parametric modulator may be able to account for more RT related variance than simply adjusting the trial duration using the RT. This, of course, does depend a little bit on the specifics. However, assuming ‘typical’ RT variations of maybe a few hundred ms from trial to trial, adjusting the length of the events will probably not result in huge changes* in the modeled timecourse after the events (which are then short boxcar functions) have been convolved with the canonical HRF, whereas a parametric regressor would give you the full variation of your RTs. And yes, a parametric regressor would also allow you to identify both types of regions: the ones that react to your stimulus in general, and the ones that have something to do with the specific response time. > > Hope this helps, > Andreas > > > *this would be in comparison to a model that uses an average RT as event duration > > > =================================================== > Andreas Finkelmeyer, Ph.D. > Institute of Neuroscience, Academic Psychiatry > Newcastle University > Campus for Ageing and Vitality, Bldg. 15 > Westgate Road > Newcastle upon Tyne > NE4 6BE, UK > > Tel.: +44 (0)191 256 3296 Fax: +44 (0)191 256 3324 > Web: www.ncl.ac.uk/ion > > > > > From: SPM (Statistical Parametric Mapping) [mailto:[log in to unmask]] On Behalf Of Andrew Jahn > Sent: 29 January 2012 19:45 > To: [log in to unmask] > Subject: [SPM] Covarying out RT > > Hi SPMers, > > I have seen a few threads on covarying out the effects of RT, and I wanted to get some clarification on the difference between modeling an event as a non-zero duration regressor, and using RT as a parametric modulator. > > As I understand it, modeling an event as a non-zero duration will account for any RT-related effects between the onset of the event and the end of the event (marked as the time when the participant makes a response). Using RT as a parametric modulator, on the other hand, will look for whether the amplitude of the BOLD response scales in any way with increases in RT. So, I presume that the approach you take depends on the question you want to answer: Whether there is a significant difference between modeling the duration of an event versus modeling it as a punctate response, or whether increasing RT leads to significant increases in the amplitude of the BOLD response. > > However, when a reviewer asks that you covary out RT (without any further clarification), is one of these approaches better than the other to address that issue, or are both equally effective ways to account for RT-related effects? > > > Thanks, > > -Andrew