The reason for this is probably a selective death of synapses in the inner ear. We separate between three types of auditory nerve fibers (ANFs): High spontaneous rate (HSR), low spontaneous rate (LSR) and mid spontaneous rate (MSR). Basically the HSRs are our high gain fibers used for low sound levels. Their response saturates quite early. LSRs on the other hand are low gain fibers used for higher levels.
the ANF synapse can release a large amount of neurotransmitter into the synaptic cleft (in this case glutamate) the problem is that high concentrations of glutamate can are actually harmful and HSR fibers seem to be quite sensitive to these high concentrations. If there is a long term exposure to a noisy environment the high glutamate concentration might selectively kills your HSR fibers. The problem with this is that your threshold of hearing is still normal as there is no damage to the HSR fibers which you need to detect low level sounds so your audiogram looks perfectly fine. But you will start to get problems if you are in situations where you need your full dynamic range e.g. understanding speech in a noisy environments.
As you can see from my writing, this is a current subject of research so we do not know for sure but it seems like that you have to be quite careful with long time noise exposure even if the sound levels are below what is normally considered as harmful.
There are some indications that people with damage might be more at risk for the hearing loss to get worse. Maybe because they where more sensitive to start with. So everyone protect your ears!!
