Looking at this thread I remembered this documentation on a room correction/eq package which talks about some research on how humans perceive tonal balance...
Link 1 Link 2
It gets pretty technical but what they're claiming is that using new psychoacoustic research they come to results which match pretty well with those old B & K frequency response targets.
The graph I attached is of the frequency response in a room at the listening position for a system that been equalized flat at the listening position using room correction. The typical room is going to cause peaks and valleys in the response which become greater in db magnitude (though not absolute energy) at the higher frequencies as shown in the graph. They are saying that human hearing pays attention more to the peaks than the valleys so the perceived frequency response is really a curve that goes along the top of all those peaks, which they draw on the graph. You can see this curve is upward sloping which means it actually sounds bright not flat. The conclusion is that room correction software needs to target a downward sloping frequency response at the listening position in order to be perceived as flat.
The curve they arrive at is flat until 100hz then slopes down evenly per octave to about -6db at 20khz, which matches the old B & K target curve pretty well.
