If I set a sat's level to be 85dB, by SPL meter, at 1K Hz and then set the sub's level to be 85dB at 40 Hz, why don't we attempt to compensate for the loudness curve and set the level higher than 85dB so the perceived level matches not the measured level?
Ideally you want your system to respond close to flat and then tilted slightly towards your room and personal tastes but still primarily so as to reproduce what is on the recording.
It is usually the job of the Mastering Engineer to decide how to balance the sound. Soft pieces tend to be mastered bass heavy to compensate for the usual fact that they are played back at modest volumes whilst heavy rock may be mixed bass light so that it can be cranked.
Examples:
Owner of a Lonely Heart - Yes - is designed to be cranked
On Broadway - Weekend in LA Live George Benson - is designed to be cranked
Killing Me Softly - The Fugees - is designed to be played more softly and has heavy bass.
Happy Coat - Osabe Trio with Ray Brown on bass - is designed to be played softly as it has heavy bass
FWIW: 100 db is about optimum flat as far as our hearing sensitivity goes and the biggest change is in the bass where extreme LF goes from being barely audible at 70 db SPL to loud at 110 db SPL.
This whole issue is clouded by the fact that most speakers compress badly when you exceed 95 db SPL at 2 meters (typical istening spot). This can ruin many otherwise excellent recordings that were meant to be cranked - as the dynamics is all lost in the BBQ temperatures inside the woofer/midrange voice coils.