Negative percent signal change

Hello

This is Muskaan from National Brain Research Center, India. I am working on an fMRI dataset acquired for the sentence reading paradigm.

I am getting a negative percentage signal change during task conditions in the language ROIs. (and also during the fixation block, though it is only 10 seconds long, which is relatively small compared to the task, a minute long).

I did preprocessing and the first and second level analysis using SPM, and for percentage signal change, I used marsbar, setting my design from first level SPM.mat file.

I confirmed the results from the code using the GUI to eliminate any chances of a methodological error.

Not all values were negative; furthermore, I see no trend for the appearance of the negative values over ROI and subject.

Random negative values appeared in visual areas as well.

Ques: Negative % signal change in the areas of interest seems counter-intuitive; does it mean deactivation?