Logo image
Exposing Pathological Sensory Predictions in Tinnitus Using Auditory Intensity Deviant Evoked Responses
Journal article   Open access   Peer reviewed

Exposing Pathological Sensory Predictions in Tinnitus Using Auditory Intensity Deviant Evoked Responses

William Sedley, Kai Alter, Phillip E Gander, Joel Berger and Timothy D Griffiths
The Journal of neuroscience, Vol.39(50), pp.10096-10103
12/11/2019
DOI: 10.1523/JNEUROSCI.1308-19.2019
PMCID: PMC6978936
PMID: 31699888
url
https://doi.org/10.1523/JNEUROSCI.1308-19.2019View
Published (Version of record) Open Access

Abstract

We tested the popular, unproven theory that tinnitus is caused by resetting of auditory predictions toward a persistent low-intensity sound. Electroencephalographic mismatch negativity responses, which quantify the violation of sensory predictions, to unattended tinnitus-like sounds were greater in response to upward than downward intensity deviants in 26 unselected chronic tinnitus subjects with normal to severely impaired hearing, and in 15 acute tinnitus subjects, but not in 26 hearing and age-matched controls ( < 0.001, receiver operator characteristic, area under the curve, 0.77), or in 20 healthy and hearing-impaired controls presented with simulated tinnitus. The findings support a prediction resetting model of tinnitus generation, and may form the basis of a convenient tinnitus biomarker, which we name Intensity Mismatch Asymmetry, which is usable across species, is quick and tolerable, and requires no training. In current models, perception is based around the generation of internal predictions of the environment, which are tested and updated using evidence from the senses. Here, we test the theory that auditory phantom perception (tinnitus) occurs when a default auditory prediction is formed to explain spontaneous activity in the subcortical pathway, rather than ignoring it as noise. We find that chronic tinnitus patients show an abnormal pattern of evoked responses to unexpectedly loud and quiet sounds that both supports this hypothesis and provides fairly accurate classification of tinnitus status at the individual subject level. This approach to objectively demonstrating the predictions underlying pathological perceptual states may also have a much wider utility, for instance, in chronic pain.
mismatch negativity electroencephalography biomarker predictive coding tinnitus phantom perception

Details

Metrics

Logo image