[pymvpa] RSA correlation > 1 ?

Nick Oosterhof n.n.oosterhof at googlemail.com
Mon Jun 13 14:22:54 UTC 2016

> On 13 Jun 2016, at 16:17, David Soto <d.soto.b at gmail.com> wrote:
> Hi, I am computing the similarity between the activation maps of 2 conditions within a frontoparietal mask using the code below,
> , and, for each individual subject. I am getting correlation values higher than 1, see output list at the bottom, however I understand these should range between 0 (identical) to 1 (uncorrelated)? 

The default metric is called 'correlation', but will actually return 1 minus the correlation. Thus the output can range from 0 (correlation=1, perfect correlation) to 2 (correlation=-1, perfect anti-correlation), with a value of 1 corresponding to correlation=0 (uncorrelated).

More information about the Pkg-ExpPsy-PyMVPA mailing list