Lindrone (and I, to some extent) already tried to explain this.
For humans to perceive sound coming from behind, the sound doesn't have to physically come from behind.
It just needs to be filtered in a similar manner as sounds that were actually coming from behind (well, at least in theory).
This filtering is based on transfer functions. Head Related Transfer Function (or HRTF) is (most often) a generic approximation of how one's torso/head/ears filter the incoming sounds from one direction.
When this function is applied to playback sounds (a game sound being played back 'from behind') based on their intended direction/position, the human hearing system assumes position based on the sound+HRTF.
Hence, we hear the sound coming from behind, even though it comes from the headphone driver usually perpendicular to the ear drum.
However, reality is not as simple as that ideal situation.
For the 'faking' of sound direction to be near ideal, the HRTFs need to be tailored individually to each listener's head. Each person has a set of (directionally based) individual HRTFs.
No game software, no sound card or consumer utility allows the measuring, uploading and tailoring of one's individual HRTFs, so that everybody can use their own HRTFs in games.
Due to this, it is impossible to get near ideal sensation of sound playback from behind. The HRTFs used in games and in sound cards (CMSS by Creative, Virtual ear by Sensaura, QSound, etc) are crude approximation: statistical averages of many different individual HRTFs. Or in some cases, even cruder approximations of human hearing related filtering, without any statistical analysis of HRTF variations.
Because these approximations are crude, the illusion of sound coming from behind is not very believable.
Also, it must be noted that human directional hearing accuracy is at it's greatest from straight behind (the cone of confusion is amongst the smallest there).
Due to this, human hearing is very adept at identifying sounds that come from behind. This is also partially the reason why it is not so easy to fool human hearing with crude HRTFs. The approximations need to be near ideal to be really convincing and stable.
So, there is nothing wrong with using 5.1 speakers or using headphones with multiple drivers.
In fact, in the first case (multiple speakers) the sound localisation can be much more accurate than with any headphones.
However, in the latter case do consider that the drivers inside 5.1 headphones really aren't positioned that much "in front" or "behind" you. They are just a little bit more angled from each other.
While this may be better than listening to single driver headphones (one driver/headphone cup) without any virtualisation in terms of positioning, it is
not as good as headphones+near-ideal HRTF in terms of positional accuracy.
A further point of interest is that there is no single set of quality 5.1 headphones (i.e. headphones with multiple drivers).
All the Zalman and other cheap copy-cat 5.1 headphones use very cheap drivers in a very passable housing. The quality leave a lot to be desired. Selection and quality in normal headphones is much better.
I'm not sure if I did a good job of explaining in a hurry, but at least I tried. Do ask if the issue is still confusing.
regards,
Halcyon
PS The most important channel in 5.1 speaker setup is center channel due to dialogue being mostly placed in that in dvd movies. This is not so important for games.
Actually, the most crucial channel in terms of accuracy in games is discrete rear channel (6.1 or 7.1 rear channel), because human hearing is most accurate regarding sound directions from behind. It is difficult to fake that position with left side rear and right side rear channels, even though center channel in front can be convincingly faked with front left and front right channels.
From this point of view, the 5.1 system was designed backwards. The center channel should have been placed behind (not in front), if one considers the human positional hearing accuracy.