I think a balanced system does both depending on the recording. The spatial information is embedded in the recording. Some mixes feature vocals or other instruments "in your face" with the rest behind and maybe lead instruments projecting forward. Others create a more homogeneous ambient sound field where everything is in the same space behind the speakers. The system/room should reveal that spatial information to reflect the intent of the artist/producer.
An example for me is Bonnie Raitt, "Blame it on Me". The vocals sit perfectly in front with the rhythm section projecting behind while the organ projects forward as if you're actually playing it.
A song like "Thing Called Love" is more homogeneous like you were watching the band on stage. The stage begins at the speakers and spreads behind.