Cortical mechanisms for tinnitus in humans /

Abstract

This work sought to characterise neurochemical and neurophysiological processes underlying tinnitus in humans. The first study involved invasive brain recordings from a neurosurgical patient, along with experimental manipulation of his tinnitus, to map the cortical system underlying his tinnitus. Widespread tinnitus-linked changes in low- and high-frequency oscillations were observed, along with inter-regional and cross-frequency patterns of communication. The second and third studies compared tinnitus patients to controls matched for age, sex and hearing loss, measuring auditory cortex spontaneous oscillations (with magnetoencephalography) and neurochemical concentrations (with magnetic resonance spectroscopy) respectively. Unlike in previous studies not controlled for hearing loss, there were no group differences in oscillatory activity attributable to tinnitus. However, there was a significant correlation between gamma oscillations (>30Hz) and hearing loss in the tinnitus group, and between delta oscillations (1-4Hz) and perceived tinnitus loudness. In the neurochemical study, tinnitus patients had significantly reduced GABA concentrations compared to matched controls, and within this group there was a positive correlation between choline concentration (potentially linked to acetylcholine and/or neuronal plasticity) and both hearing loss, and subjective tinnitus intensity and distress. In light of present and previous findings, tinnitus may be best explained by a predictive coding model of perception, which was tested in the final experiment. This directly controlled the three main quantities comprising predictive coding models, and found that delta/theta/alpha oscillations (1-12Hz) encoded the precision of predictions, beta oscillations (12-30Hz) encoded changes to predictions, and gamma oscillations represented surprise (unexpectedness of stimuli based on predictions). The work concludes with a predictive coding model of tinnitus that builds upon the present findings and settles unresolved paradoxes in the literature. In this, precursor processes (in varying combinations) synergise to increase the precision associated with spontaneous activity in the auditory pathway to the point where it overrides higher predictions of ‘silence’.