Why can't I just duplicate and pan a sound?

[cron]

Just to tie mine and planetearth's replies to your question together, although it looks like we're describing very different things, we're describing the same effect. Leaving the relative levels of mid/side untouched means you get a perfect recreation of the left/right signal. The side channel alone when soloed however is always 180 degrees out of phase when converted back to left/right, meaning that it'll completely cancel itself out if mixed to mono. As you widen your signal by boosting the side relative to the mid, the more messed up the phase becomes, and thus the more problems you can expect with mono compatibility.

Although it isn't technically correct, it can be useful to think of the mid as being 'the things left and right have in common' and the side as being 'the differences between left and right.'

[Juljan]

Thanks once again for your help on the widening question.
Another short one: How is it possible to trick the mind that you hear sounds from the rear side of your headphones?

[cron]

The short version is that the externally visible parts of your ears are a very complex shape/structure for this reason. In the real world, depending on the direction the sound comes from, the sound will be filtered by the shape of your ears and its varying thicknesses differently (and also by the shape/consistency of your head and its innards - particularly important if the sound arrives at each ear at slightly different times). This is how we can figure out exactly where sound is coming from in the real world, despite only having two ears. Binaural recording works on this principle. In the old days, you'd build a very accurate head with very accurate ears out of some material that reasonably approximates flesh, and shove microphones into its ears. The result is that the sound would already be 'prefiltered' by the time it reached the microphones, and so these directional cues would be embedded into the recording. For obvious reasons, the resulting recordings work best with headphones so that the listener's head and ear shapes are (to a large extent) removed from the equation. On headphones, the sound is being piped directly into the ear canal and so the listener is only hearing the filtering from the dummy head/ears that the recording was made on. These days, no dummy head is required and we can do it in plugin format with some variation on the HRTF (head related transfer function) principle.

[planetearth]

To follow on with what @cron was saying, you're getting pretty deep into psychoacoustics/sound localization when you start trying to "fool" the listener by trying to make it appear that a sound is coming from behind (or above, or below) him if you're only using two speakers.

Basically, directional cues (the hints that tell us where a sound is coming from) are in the higher frequencies; low frequencies don't appear to come from any particular direction, which is why you can put a subwoofer pretty much anywhere. But the higher frequencies arrive at the ears at different times, and that tells the brain where the sound is coming from. And this doesn't work on everyone, since we all don't hear high frequencies the same way. But this is why the headphone trick @cron mentioned works: the ears are hearing high frequencies at different times (and sometimes, through the flesh of the ear or the head itself), and that's what makes a person think a sound is behind him.

You can avoid the hassle of recording your audio "binaurally" (with a "dummy" head and two matched microphones in the ears), and you can try to do this with a plug-in, but you may not get the results you want. The amount of high-frequency information in your audio, the distance "behind" the listener you're trying to achieve, and the effectiveness of the plug-in will all be contributing factors. That said, if I were going to try this, I'd reach for a plug-in, too. I just don't know which one, because I've never tried it.

You may also want to keep in mind that any "3-D" effect you get with a plug-in in headphones will almost certainly not work over speakers. You need the headphones to control how the localization information/high frequencies gets to the ears. If you play the same audio over speakers in a room, signals from the right speaker will bounce off walls and objects around the listener, and arrive at the left ear (and vice-versa). This will destroy the localization cues you built into the audio with the plug-in.

Hope that helps!

Steve