Dear everyone,

   I'm an undergraduate student of Peking University, China. I've recently done an fMRI experiment focusing on human sequential processing.    I have tried a  model in analysing my fMRI data and got rather appealing outcomes. However, i'm not sure about the reliability of the analytical method  we used. We hope that you can give me some advice.
   I use Chinese characters as visual stimuli. If you have ever watched someone writing Chinese, you may know that Chinese charaters consist of several strokes, which form a 2D spatial structure. When we Chinese people write characters, we follow a specific sequential order that desinates which stroke is to be written first and which is to be followed and so on. This sequential rule is encoded in the early stage of literate individuals when they first learn to read and write the characters and thus is an well-learned, embedded sequential representation.
   In our experiment, we presented to the participants the dynamic writing process of Chinese characters, one stroke after another, covering on an framework of the specific charater being written. Some of whose strokes order is correct and some  is violatedŁ¨By violated, i mean, reverse the presentation order of two consecutive strokesŁ©. The design matrix we used contained regressors of the onset of the beginning of each trials in each condition(correct and violated), the onset of the violated stimilus in violated condition and the corresponding stimulus in ordered condition. Additionally, there are also regressors containing the onset of participant responses and behaviously incorrect trials.(since these two regressors are unimportant, we'd better forget them hereafter and keep in mind only two regressors concerning the onset of the beginning of the whole trial and the onset of the critical stimulus) As you can see, by so modeling, i have divided each trial into two parts: one before the critical stimulus and the other after the critical stimulus. As for the duration, i set them equal to the real duration in the experiment. Then in the contrast stage, we combined the two parts of each condition to form a whole(in terms of SPM, we set the two regressors in one condition to be 1 and those in the other condition to be -1  ) and contrast them with each other.  i'm not sure whether this method is feasible, since i haven't introduce jitters before the critical stimulus.

 

Thank you.