Probability theory is applied to the analysis of fMRI data. The poster
ior distribution of the parameters is shown to incorporate all the inf
ormation available from the data, the hypotheses, and the prior inform
ation. Under appropriate simplifying conditions, the theory reduces to
the standard statistical test, including the general linear model. Th
e theory is particularly suited to handle the spatial variations in th
e noise present in fMRI, allowing the comparison of activated voxels t
hat have different, and unknown, noise. The theory also explicitly inc
ludes prior information, which is shown to be critical in the attainme
nt of reliable activation maps.