This small extract of the "Yellowstone" nature program from the BBC shows a fox which is using his sound localization ability to hunt prey animals that are concealed under a cover of snow.

This figure shows acoustic cues to sound source direction. It is a color version of Fig. 5-2 of "Auditory Neuroscience", and is based on acoustic recordings from my own ears carried out by Prof Doris Kistler at the University of Wisconsin at Madison.

The data are shown as spherical map projections, with my head at the centre of the sphere, and the interaural cue values shown by the color codes for the corresponding sound source directions in azimuth and elevation. A and B show interaural level differences (ILDs) in decibel. As you can see, ILDs are highly frequency dependent, i.e. high frequencies (say near 11 kHz, panel B) can generate large ILDs of 30 dB or larger, while lower frequencies (say near 700 Hz, panel A) generate ILDs of well below 10 dB. 

ILDs are also significantly affected by the geometry of the head, outer ears, and shoulders, which is why the patterns of the ILDs are much more irregular than the Interaural time difference (ITD) patterns shown in C. ITDs are nearly spherically symmetric around the interaural axis, are much less frequency dependent, are maximal for positions right off to one side of the head, and never exceed values of 700 microseconds or so.

ITDs and ILDs are both valuable cues for localizing sound sources, but even individuals who are deaf in one ear can sometimes localize sounds in space with some degree of accuracy. These judgements are thought to be based on monaural spectral cues (shown in D for sound source locations right in front, 0 degrees azimuth, at various elevation). Spectral cues arise because the outer ear filters sounds in a manner that depends on the angle of incidence of the sound waves, creating patterns of peaks and notches in the high frequency end of the sound.

The Jeffress Model has long been a popular model for trying to exlpain how the mammalian Medial Superior Olive (MSO) or the bird nucleus laminaris may extract interaural time differences for sound localization. The Jeffress model posits a "delay line and coincidence detector" arrangement which is illustrated in the animation shown here. The animation nicely conveys the appeal of the Jeffress model. Due to a systematic delay line arrangement and a requirement for precise synchrony in the activation of MSO neurons, different parts of the MSO become sensitive to sounds from particular directions. But note that the cartoon shown here is in many ways "biologically naive", oversimplifying the anatomy and physiology. Whether the Jeffress model, such as it is shown here, is really a good description of the operation of the mammalian MSO is becoming increasingly controversial.
(Note: the video contains no sound).

Acknowledgement: this video is from the web page of the laboratory of Prof. Tom Yin of the University of Wisconsin, one of the pioneers of research on the physiology of the mammalian MSO.

Sound localization relies heavily on interaural disparities (i.e. differences in the signals received in the left and right ears), and these differences are larger, making sound localization easier, if the two ears further apart.
Devices which artificially separate the separation between the ears, for example with suitable tubing, can therefore make it easier to pinpoint the direction of sound sources. An impressive example of such a device, the "hearing throne" of the Oldenburg Hearing-Gardens, is pictured here:

Also, if the ears are offset vertically rather than aligned horizontally, this may help judge the elevation of a sound source. The following images, from the collection of the Waalsdorp Museeum in the Netherlands, are from an era when radar was not yet widely available, and show attempts to use this fact to develop devices to locate enemy aircraft:

Further examples of devices intended to improve spatial hearing, including this wonderful portable contraption, can be found at http://www.damninteresting.com/can-you-hear-me-now

This page has little animations illustrating the two major binarual cues for sound source direction: Interaural Time Differences (ITDs) and Interaural Level Differences (ILDs).

You will need to listen to the sound tracks of the videos using headphones. To perceive the effects, you need to have good hearing in both ears. If you have a temporary or permanent hearing loss in one ear, the demos won't work for you. (sorry!) They also may not work if your headphones or the sound card on your computer are poor quality. A number of laptop sound cards in particular "cut corners" when it comes to reproducing some temporal features of sound with great accuracy, which will cause these demos not to work correctly. Note also that many wireless bluetooth headphones are not regenerating ITDs properly, so if you use wireless headphones the ITD part of the demo may not work properly either.  Also if you have the headphone the "wrong way 'round" (i.e. the transducer that is meant to go to the left ear connected to your right ear) then the sound may appear to come from "the wrong side". 

What to observe

The demos below show and play short 500 Hz tone pips which vary either in ITD only, or in ILD only, or in both ILD and ITD at the same time.

The first demo shows variation in ITD alone, starting with the left ear leading by 0.4 ms, then the ITDs shift in steps of 0.2 ms until the right ear leads by 0.4 ms, and then they shift back. If the demo is working as it should, you will hear the sound source appear to move from a place somewhere slightly to the left to somewhere slightly to the right.

Varying ITDs Only

The second demo keeps ITDs constant at zero, but changes ILDs, so that the left ear sound is initially 6 dB more intense. The ILD then shifts in 3 dB steps to the right, and back again. These ILDs are exploited in stereophonic music, and you will not be surprised that they can appear to shift the sound source to the left or to the right, but you may find it peculiar that either changing ITDs or changing ILDs can lead to similar perceived changes in sound source position, even though they do very different things to the sound.

Varying ILDs Only

Of course, for normal, free-field sound sources, ITDs and ILDs covary, i.e. the sound will be both earlier and more intense in the nearer ear. The most natural situation is therefore the one shown in the 3rd demo, where ITDs and ILDs vary together. If your ears are like mine, then the impression of a moving sound source in this 3rd example will be clearer and more compelling, and the source will seem to move over a wider range, than in the first two examples.

Varying ITDs and ILDs together

With artificial headphone sounds we can also make ITD cues and ILD cues oppose each other, i.e. the sound might start earlier in the left ear but be louder in the right. With such conflicting cues, our brains tend to perceive "compromise" positions near the midline, a phenomenon described as "cue trading". This is illustrated in the fourth example. Here, the sound source should appear to move much less than in the 3rd, and possibly also less than in either the 1st and 2nd examples. 

Trading ITDs off against ILDs

Individuals tend to vary somewhat in how sensitive they are to ITDs, and if your sensitivity to ITDs is very low then you may not hear any movement in the first example, and not much difference in the 2nd to 4th examples. (However, it is also possible that the sound card or sound software on your computer is cutting corners and not reproducing ITDs accurately, as seems to be the case for example on my Acer Aspire 1810 laptop). In contrast, if you are very sensitive to ITDs, you may find that the first example gives more compelling movement than the 2nd. Your individual sensitivity to ITDs will affect how much movement you hear in the 4th example, if any, and in which direction.

Here below you can find the Matlab source code for a program authored by Matthieu Lesburgueres and Jan Schnupp which you can use to run a psychoacoustic experiment to measure your own sensitivity to ITDs and ILDs.

Move the mouse cursor over the grid below to hear a harmonics complex tone with ILDs and ITDs as indicated.

You should listen to these sounds over headphones, and this demo may not work for you if you have a significant hearing loss in one ear, or if the sound card on your computer is of poor quality and does not separate the two stereo channels well. Note also that many wireless bluetooth headphones are not regenerating ITDs properly, so if you use wireless headphones this may not work either. 

Positive ILDs and ITDs should make the sound appear to come from the right. Hence, the sounds toward the top right should sound the furthest right, the ones in the bottom left the furthest left. (Make sure you have your headphones on the right way around.)

The curious thing to observe here is that, if you start from the middle, you can move the sound right either by moving the mouse cursor up, or by moving it right. You change the sounds in two very different ways (in one case you change the timing between the ears, in the other you change the relative sound intensity) but although the manipulations are very different in nature, they produce the same effect: the sound appears to come from the right. Can you tell by listening carefully whether the sound percept moves because of timing or intensity cues?

Another thing to try is this: start from the middle, move the sound right by moving the cursor right, then move the cursor down. As you move the cursor down the sound should appear to move back toward the middle, because the timing and intensity cues now point in opposite directions, and your brain perceives the sound at a compromise position somewhere in between.

As discussed in the context of Figure 5-5 of "Auditory Neuroscience" (reproduced here below) ITD cues for sound location are derived from the interaural phase. Conequently, tones that are slightly mis-tuned and which are delivered separately to the left and right ear can give the impression of a shifting lateralization from left or right. THe sound example here below shows this. To the left ear we play a 500 Hz pure tone, to the right ear a 500.25 Hz pure tone. So the left and right ear go in and out of phase once every 4 seconds. The left and right ear start in phase, so when listened to over headphones, the sound should start off sounding as if it was in the middle. However, since the frequency in the right ear is ever so slightly higher (the oscillations are ever so slightly faster), over the next 2 seconds the right ear's phase starts to lead by up to 1 ms, giving a changing ITD cue that suggests that the sound is shifted to the right. After more than 2 seconds, the right ear phase leads by more than 1 ms, but since the period of the tones is 2 ms long, the brain may interpret this as a less than 1 ms phase lead in the left ear. So about 2 seconds into the demo, the sound will sound as if it is now suddenly coming from the left, but then it will gradually shift over to the right again over the next 4 seconds, only to then jump again to the left, and so forth.

Steady pure tones are unpleasant to listen to, and they also produce a great deal of adaptation in the auditory nervous system, making it difficult to perceive them well. We therefore added a 5 Hz sinusoidal amplitude modulation to the this demo. It makes the effect easier to hear. 

This demo is designed to be listened to over stereo headphones. It may not work over loudspeakers. Also, some individuals will have difficulty hearing the effect, particularly if they have a history of ear problems in either ear which may interfere with binaural hearing.

The numbered squares signify the sound directions corresponding to a series of "virtual acoustic space" stimuli, which were generated by convolving a stimulus - in this case, a series of tapping sounds - with the head-related transfer function of one of the authors of this book.

Put on a pair of - preferably fairly good - headphones and plug them into the headphone socket of your computer. By moving the mouse over the squares, you will hear the stimulus changing location. Most stererophonic recordings are based only on differences in level between the two ears and the resulting sounds are therefore lateralized toward one ear or the other. By contrast, virtual acoustic space stimuli incorporate the full complement of localization cues and, in principle, replicate real free-field sounds. Consequently, as you move the mouse, you should hear the sounds move up or down as well as left or right and even seem as though they are either in front of or behind you. Of course, how well this will work depends to some extent on how closely your head and ears match those of the subject from whom the acoustical measurements were made to generate these stimuli.