There's a lot of work on how the neocortex is special in implementing computations that subcortical structures are (presumably) not suitable for.
But it seems birds achieve the same outcome with a nuclear organization. So why not the mammalian subcortex as well?