The effect size can be consistently negative across conditions if the BOLD
signal in the ROI is less than that of the whole brain mean. If you've
modeled the fixation separately, then you may have a rank-deficient design
matrix in which case the individual PE values aren't entirely meaningful --
what is important is the relative differences between PEs. It sounds like
your contrast is giving you meaningful results consistent with your
expectations, so I wouldn't worry about the absolute value of the
Hope this helps.
Joseph T. Devlin, Ph. D.
FMRIB Centre, Dept. of Clinical Neurology
University of Oxford
John Radcliffe Hospital
Headley Way, Headington
Oxford OX3 9DU
Phone: 01865 222 738
Email: [log in to unmask]