I don't understand the question.   Once you have looked at cochlear microphonics, auditory nerve and brainstem potentials, evoked cortical responses, etc, what more information do you want?  These tests depend on averaging, which would not work unless the responses were stereotyped.

Konstantina Stankovic at Otology and Neurology, Massachusetts Eye and Ear Infirmary has published work on cochlear implants and may know the answer.

It is simple when you describe it this way -  neurons fire pulses, and more of them if more agitated. A cochlea is a frequency analysing device, so pulses are related to the frequency content of the sound. Things tend to complicate when you go to details, but in its basics the process is beautifully simple.

The cochlea's role in frequency analysis is seen from a pitch space, and the way we perceive music.

By studying animal sonars I may conclude that mammal hearing is inherited from a common ancestor that was a nocturnal animal relying on sonar akin to the modern bats. It has a perfect equaliser response curve, fine perception of frequency, and a perfect distortion for demodulation purposes.

@Veljko In 5 weeks  I'll be passing through Boston and may pay Ms. Stankovic a visit, thanks!

@Davor I know that things might get complicated but that is why I asked the simplest possible question. As far as I understand, neurons process information by receiving and sending neuronal pulses which are more or less of standard amplitude and width so that it is believed that only their temporal order carries information. It therefore must be that something in the ear (cochlear or something after it) generates neuronal pulses (on one or many channels) that are further analyzed/processed by the brain. My question is how the train of neuronal pulses evoked by a single tone looks like. And what changes of the sound intensity jumps by a factor of 2?

There is a lot information on pitch mapped neurons firing mechanisms in papers on cochlear implants. Say Philipos C. Loizou's http://ecs.utdallas.edu/loizou/cimplants/chap_loizou_review2006.pdf

There is a chapter on neuron firing pulse repetition rates (frequencies).

Hi Mario,

You might find it worth your while to buy a good book on hearing, such as James Pickles An Introduction to the Physiology of Hearing, Academic Press. You can sometimes get an older edition second hand.

The cochlea of mammals contains two types of sensory hair cells. One type, the outer hair cells, have almost no nerve connections to the brain and are mainly involved in amplifying incoming sounds and driving the movements of th ebasilar membrane that underlies al lthe cells of the organ of Corti. The second kind of sensory hair cells, the inenr hair cells, have many (up to 30) afferent fibers connected to them. They belong to several classes, which differ in their sensitivity. Thus there is one class that is highly sensitive and in the presence of low-level sounds or tones of their preferred frequency send action potentials to the brain. The number of action potentials per second increases as the sound level increases. At somewhat higher levels of sound other, less sensitive nerve fibers begin to discharge and their rate of firing also increases with sound level. Thus the different sound levels are coded both by (a) which classes of fibers are firing and (b) what is the firing rate of each fiber class.

The firing pattern of the fibers can also code other information, such as whether a tone is continuous or not. Thus a complex sound such as speech is coded by a variety of fibers arrayed along the hearing organ, each able to code level and timing information for its own frequency response range. Finally, at low frequencies (in humans below about 2-3 kHz), the fibers are able to arrange their firing pattern in such a way that each fiber preferentially fires at one phase of the sound. This is called phase locking and, through analysis of the duration of the time period between spikes, helps the brain to determine the sound frequency.

In the brainstem there are several so-called auditory nuclei that in parallel receive input from auditory nerve fibers. These nuclei are large masses of neurons specialized to process the input with regard to their own "special" feature, such as sound level. All this complex analysis is then combined at higher levels, eventually allowing a "percept" of the sound to occur. These patterns of neural coding - with the exception of the two types of hair cells - are found in all land vertebrate hearing systems.

Best regards,

Geoff Manley

Marlo, there are a couple of things to take into account here. 1. The activity at the 1 kHz place in the basilar membrane will be activated at the lowest sound pressure levels. The activity of auditory nerve fibers connected to this spot show increased number of action potentials at the lowest sound pressure levels, and the number of action potentials increase. Potentially, the largest increase in the number of spikes fired is at ANFs connected to this location.

2. There is activity in adjacent basilar membrane locations, but the activity decreases very rapidly on the high frequency end, and slowly at the low frequency end. So, there is a spread of activity among ANFs tuned to neighboring frequencies in a manner similar to basilar membrane activation.

3. There is also going to be some amount of activity that is temporally synchronized with the stimulus (phase locking). Ass the above poster said, the temporal pattern of activity is rearranged so it is synchronous with the stimulus waveform. Since this is a human, and we have very little data on phase locking in humans, we extrapolate from animal models. The 1 kHz place shows good phase locking in animal models. So, we predict there will be phase locked activity, locked to the 1 kHz frequency.

4. This is transmitted to the ascending auditory pathway via the cochlear nucleus, the superior olive, etc. There is a complex series of interactions at all of these locations and feature extraction takes place. Result: some higher order neurons show increased firing at your condition, some show decreased firing. There is some amount of phase locking, dependent on the site you consider. All of these are somehow decoded (this is where you say some magic happens), and you perceive the intensity and frequency of sound.

In summary, it is the pattern of activity over time (both the number of spikes and the temproal pattern of those) that will determine perception.

SInce you mentioned being in the Boston area, you should also consider talking with Bertrand Delgutte, Dan Polley, Barbara Shinn-Cunningham, and Charlie Liberman in addition to Constantina Stankovich.

Thanks to all.

@Geoffrey. OK lets now assume constant sound level and look at the affected hair and its 30 afferent fibers. Let us further concentrate on one afferent fiber that is sensitive in the given sound level range. This fiber generates action pulses at a given, constant rate as long as the sound level is constant, right? Say 100 pulses per second, that is one pulse every 10ms on average. My question is how this sequence of pulses looks in time?  Are all pulses exactly equally spaced in time, that is one pulse exactly every 10 ms? Or spacing between subsequent pulses varies but it is equal to 10 ms when averaged over long time? Is it possible to find in the literature an actual recording of these pulses? Note that this is not an academic question but can have profound consequences.

By the way, what is the minimum and maximum action potential firing frequency of one afferent fiber? Does this vary with the frequency to which the hair is sensitive to or do they all have the same firing capacity?  

@Ramnarayan. I have a book on HiFi that says that human hearing is "insensitive to the phase", which would be true if cochlear would be just an FFT spectral device, but I always suspected this is not true and that a good deal of 3D location by sound comes from phase information. For example, if you are unscrewing a small screw near your face and it suddenly drops to the floor, since it is small and eyes cannot locate it and focus quickly enough, you will have much better chance to locate where it went by concentrating on sound of it bouncing on the floor. I experienced this a 1000 times. Since sound waves are long, it is fascinating that a small object can be located with a few centimetre precision by ears, and this can only be explained by resolving the sound phase.  Now you seem to confirm that. However, at the moment this interesting question  is out of my scope

Mario,

In the firing for shorter time periods, the rate varies stochastically, even if an auditory fibre is phase locked to a high degree (perfect phase locking is not observed, but vector strengths can be very high, meaning that the vast majority of intervals are locked to the phase of the stimulus). So it really is a question of what you mean by constant. The most exact spacing in time is only achieved during phase locking to high-level sounds at low frequencies – that is frequencies below the maximum firing rate of the fibre. Above that frequency, some of the cycles of the stimulus will be “missed”, i.e. the fibre will not be able to fire again that quickly. However, the spacing of the next firing will still be closely related to the phase (one, two, three cycles, etc). There is no minimum firing rate of auditory neurons in the cochlea – all of them have spontaneous activity that can be from below 1 spike/s up to about 100/s. During phase locking, such spontaneous firings become “locked” to the sound stimulus. The maximum firing rate of primary auditory neurons is somewhat higher than 300/second, pretty much irrespective of the best-response sound frequency. This maximum rate is not restricted to the auditory system but is true for all neurons of endothermic vertebrates. The longer the sound pulse, the more the rate will fall over time, often reaching a lower plateau rate after about 50 -100 ms (this is called adaptation). As I wrote before, it is likely that at low frequencies, the brain uses the spacing of the firings to derive the frequency of the sound (originally called the “volley theory”). Above frequencies where the spacing between firings no longer contains this information, i.e., above perhaps 1-2 kHz in humans (this is a guess), the information concerning the frequency of the sound is only conveyed by the place of origin of the fibres in the cochlea (tonotopic organization). What a firing neuron sounds like (the one in this video has a very variable rate and is not, I think, an auditory neuron) can be heard at:

https://www.youtube.com/watch?v=8bxpz-YEuao

I am not aware of a web site that offers a real sound recording of cochlear neurons firing, but some firing patterns are shown, for example in:

http://www.neurophys.wisc.edu/h&b/textbook/chap-6.html

Geoff Manley