Tinnitus in a computer model

Tinnitus in a computer model

Editor: As something of a technogeek, I think this is a very interesting development, and I’m hopeful that it will increase our knowledge of tinnitus and how to overcome it. But I’m also mindful of the old computer science slogan of “garbage in, garbage out”, which means that the results of the computer model are only as good as the assumptions the modelers made.

Scientists from Berlin study how hearing loss can lead to tinnitus
Tinnitus, i.e. the perception of phantom sounds in the absence of an acoustic stimulus, can be caused by hearing loss. Under which circumstances does this occur? Which mechanisms are involved? Roland Schaette and Richard Kempter from the Bernstein Center for Computational Neuroscience and the Humboldt University in Berlin found answers to these questions using computer simulations.

Tinnitus arises in the auditory pathway of the central nervous system. In animal studies, tinnitus-like activity of neurons – so-called hyperactivity – has been found in the dorsal cochlear nucleus (DCN), the first processing stage for acoustic information in the brain. Neurons of the DCN receive input directly from the auditory nerve and react to it with neuronal discharges – one says, they “fire”. Even without any acoustic signals, however, cells of the auditory nerve and the auditory pathway are still active and fire spontaneously at a certain rate, the “spontaneous firing rate” – comparable to the background noise produced by electrical devices. Various studies suggest that hearing loss can increase the spontaneous firing rate of nerve cells in the DCN and that animals perceive this as a kind of tinnitus. In a theoretical model, Schaette and Kempter explain the link between tinnitus and hearing loss for the first time.

After hearing loss, auditory nerve fibers and neurons along the auditory pathway only react to loud sounds. For soft sounds below the increased hearing threshold, the neurons fire spontaneously. Many neurons thus show an overall reduced activity. This could trigger a mechanism called “homeostatic plasticity”, which ensures that neuronal activity is neither too high nor too low. If the average activity of the neurons is too low, homeostasis enhances their sensitivity. As the scientists could show in their model, neurons then react more strongly to the activity of the auditory nerve; in particular the spontaneous firing rate increases.

Moreover, Schaette and Kempter also demonstrated in their model that this mechanism only applies to certain types of neurons – for example to type III neurons of the DCN. These neurons are primarily activated by sound. Therefore, their average activity initially drops after hearing loss and the mechanism described above is initiated: homeostasis has to counteract this loss in activity and elevate firing rates, which then also leads to an increased spontaneous firing rate. In contrast, type IV neurons are either activated or inhibited by sound, depending on sound intensity. Hearing loss only has a minor effect on their average activity. Accordingly, these neurons are less susceptible to hyperactivity. This prediction of the Berlin scientists’ model corresponds with experimental findings: In rodents type III neurons dominate in the DCN. Here, tinnitus-like hyperactivity has been observed. In contrast, such an activity has not yet been found in cats, whose DCN mainly holds type IV neurons.

“Our studies have corroborated the association between hearing loss and tinnitus, which could provide a foundation for new treatment strategies,” Kempter states. “Our hope would be that a tailored exposure to acoustic signals over an appropriate frequency range could help to drive back the hyperactivity caused by hearing loss”.

Original publication:
Schaette R, Kempter R: Development of tinnitus-related neuronal hyperactivity through homeostatic plasticity after hearing loss: a computational model. Europ J Neurosci 23:3124-38 (2006). doi: 10.1111/j.1460-9568.2006.04774.x

Schaette R, Kempter R: Development of hyperactivity after hearing loss in a computational model of the dorsal cochlear nucleus depends on neuron response type. Hear Res 240:57-72 (2008). doi:10.1016/j.heares.2008.02.006