Hi all,
I'm trying out the multivariate bayes method in SPM comparing several models of neural coding, however I'm not very clear on the meanings of the various model priors. The spm_mvb_U script lists the priors as follows:
priors - 'null' % no patterns
- 'compact' % reduced (ns/2); using SVD on local compact support
- 'sparse' % a pattern is a voxel
- 'smooth' % patterns are local Gaussian kernels
- 'singular' % patterns are global singular vectors
- 'support' % the patterns are the images
Could someone please either explain what these mean or point me to a resource that will explain them? For example, what does 'a pattern is a voxel', or 'patterns are local gaussian kernels' mean? I've read the Friston et al paper in NeuroImage on the method, but it was very technical and I didn't really understand most of the maths.
Thanks in advance,
Ciara
|