Engineers have developed digital signal processing methods built on these psychoacoustic principles which are perfectly capable of fooling the human brain into perceiving multiple sounds coming from locations other than the speakers from which they're really emanating. Again nothing extremely fancy in terms of the concept is being done... just alterations to frequency and amplitude which produces the phase characteristics that make sounds appear to be projecting from different locations... and not merely one complex sound at a time, but many complex sounds at a time.
There are products out there that can fool most people...most, not all. Expert virtuoso players can usually pick out sampled and synthesized instruments from real ones. But even with digital replication and modeling of physical sound, I can assure you that most audio and music industry professionals (for the most part those who actually do more than research)
will be able tell the difference between any modeled (to a "T" even) instrument @ AAC 256kbps and a true instrument recorded @ 24-bit 96KHz PCM. And I am confident most would prefer the "true" track. Total speculation I understand, but to live solely in the realm of numbers and whitepapers can really put one out of touch with practical reality.
The main reason digitally synthesized and sampled instruments (such as Synthogy's Ivory, for instance) sound so good is that they were recorded at very high bitrates and bit depths. You would have to be crazy to or just stupid to believe that 256kbps audio is good enough for even the crappiest multitrack demo recordings.
If you ask most digital photographers, what it is that contributes the most to their image quality, they will often list optics the highest, then CCD quality, and then megapixel count. And while, sure a 3MP picture through a 2500 dollar lens will probably look pretty damn good compared to a 8MP picture shot through the same lens downsampled to 3MP, perhaps even indistinguishable, the fact of the matter remains, @ 3MP you are stuck @ 3MP. You will never be able to retrieve that data back. While the eye might be more sensitive to dynamic differences than the ear, the analogy ports over ok. A 256kbps track using today's encoder's will never sound as "good" as a PROPERLY gain staged and mixed hi resolution recording, despite today's advanced codecs.
Sadly though, the over-discussed "loudness wars" have degraded the quality and dynamic range of most commercially available recordings to the point that yes, that Madonna track bought off iTunes is likely completely indistinguishable from the CD cut, even on the highest of hi-fi component systems.
Through your iPod headset, forget it. Youll never tell the difference in a thousand years of blind testing.