Our brains subconsciously discount reverberation unless the sound is reversed

Evidence

Playing sound backward that was recorded earlier in a busy room is revealing. The frequency content is the same, and the backward sonogram looks very similar to the ordinary one, just reversed. But separating a voice from background noise becomes difficult, as does recognizing the sound of a dish breaking as opposed to a metal tray being dropped in a crowded restaurant. Indeed, as we have already noted, the reverberations that are so constantly with us are duly noted subconsciously, until the sound file is played backward. Then, the true duration of the reverberations reveal themselves. — Eric J. Heller, Why You Hear What You Hear- An…, 2530

Categorization

• QUESTION: What exactly is lost when reversing audio that makes voice separation so much harder? Is it purely the forward temporal pattern of reverberation decay, or something subtler about attack transients?
• XREF: Connects to reverse-speech phenomena and auditory scene analysis (Bregman’s Auditory Scene Analysis) — the same mechanisms of stream segregation.