Insulating BC/orientation of normal

[kirip]

I am using the Static Current solver and have attached pic for geometry (outer boundaries are further out and potential source is a dipole, but am mainly interested at what is going on at this finer scale).
Briefly, I am trying to model electroreceptive organs of sharks and have two materials, one with the conductivity seawater and one which is a tenth of that. I think the meshing etc works for that, but now wanted to implement a BC saying that the walls of the organ (indicated in red) are insulators. I tried to do this by setting the a Neumann BC with current density=0 but when I calculate the Efield and plot it doesn't seem to work (see pic 2). I did not expect to see the EField (black line with arrows) penetrating throught the walls.
At a guess I would say this is something to do with how the normal is orientated (i.e. inner vs. outer or directed towards higher conductivity)? Does anyone know if it is possible to specify its orientation for the static current solver, or do it prior to elmer either in gmsh or ElmerGrid?
Thanks Kiri