Encode data-dependency in a distribution

Dear all,

I’m new here. If possible, I would like to encode data-dependency in a distribution. For example, if my datasites are located in the interval [0,10], I would like to set the prior of the standard deviation in a certain way in [0,4] and differently in [4,10].
Is this possible? If yes, could you provide a really simple and brief example for the implementation?

I read about the Bound construct, but I don’t understand if it could be helpful in this situation.

Thanks a lot