Some types of artificial sounds presented through headphones are sometimes described as not natural, in the sense that they have binaural relationships that sounds in a natural environment do not have. In general, this qualification refers to the qualities of point sources in an anechoic environment, but real environments reflect sounds and there are also more complex sound sources. I will discuss two types of “unnatural” binaural sounds.
1) Binaural noise with long interaural delays (ITD). In an anechoic environment, the ITD of a sound can reach 600-700 µs in high frequency for humans, and perhaps up to 800-900 µs in low frequency. Yet if we listen to binaurally delayed noise through headphones with an ITD of about 10 ms, we hear a single source, lateralized to one side. When the ITD is increased, starting from 0 µs, perceived lateralization progressively increases up to about 1 ms or a bit less, then reaches a plateau. We hear two separate noises only when the ITD is larger than about 10 ms. This is surprising because 10 ms is much larger than the maximal ITD that can be produced by a single sound source in an anechoic environment. However, let us consider a situation where there is an acoustically reflecting surface, a vertical wall, which is on our left side, about two meters away. A sound source is far on the opposite side. In this case, the right ear receives the direct wave from the source and the left ear receives the reflected wave. It follows that the ITD is about 10 ms. In addition, the direction of the sound source is consistent with the headphone experiments. Therefore, large ITDs may not be unlikely in natural environments, even with a simple point sound source.
2) Uncorrelated binaural noise. If acoustical noise is made of the addition of independent many sound sources, one would expect that the signals at the two ears are correlated in low frequency, that is, when the period is large compared to the maximum ITD of the sound sources. Are there sounds in an ecological environment that are binaurally uncorrelated? I would suggest the following situation. You are riding a bicycle, and you feel the wind in your face and in your ears. The sound of the wind in the ears does not feel localized anywhere else than at the ears. There is also little reason to believe that the pressure of the air is highly correlated at the two ears, except perhaps at very low frequency - although this should be measured. In fact, this acoustical situation correlates with mechanical pressure on the ears, which can be captured by tactile receptors. I would suggest that this type of sound is perceptually localized at the ears because of this ecological association.
In summary, acoustical situations are considerably diverse in ecological environments, and therefore there might be fewer “unnatural” sounds than often assumed.