Permutation in RSA Analysis

Hi guys,

I am conducting an representational similarity analysis on CosmoMVPA. I have 4 conditions, so 4 types of targets. Everything seems to go fine until the permutation test.
When I conduct a simple randomization of targets and run the searchlight again, I get exactly the same results. I tried with different masks, different procedures (volume-based, surface-based), when I try the permutation test, the results are respectively the same. Has anyone ever experienced that? If so, where could be the problem?

Thanks in advance