+1 to DSOTM in Atmos. I manage 4.4.1
In real life, the shape of the pinnae surrounding our auditory canal plays a critical role in detecting sounds coming from behind. It is our principal method of detecting danger.
With two-channel stereo, if the signal contains corresponding information, you will localise the source. Chesky did this with their LEDR test where the apparent source rises vertically from one speaker, arches over before descending towards the other speaker.
The more coherent the speaker, the easier the effect is to hear. The information is mainly carried by the high frequency components of the signal, so clean high frequency response helps.
As we age, we lose high frequency 'reception' so a combination of deafness, old age, in-coherent speaker design and confusing reflections combine to mask the effect

