Cortical determinants of human auditory cognition

Hearing loss due to ear damage eventually affects half the population, is a major obstacle to work and home life, and a drain on the UK economy. If you ask anyone with common hearing loss what is wrong, they will NOT say 'I have lost 40 decibels at 2 kilohertz': the hearing-test result. They WILL say that they cannot hear colleagues at work in the canteen, or their family at dinner. Listening to speech in noise is much more complicated than the simple detection of sounds measured with hearing tests. We have to separate overlapping speech sounds and background noise that are louder than the sounds in hearing tests, group together speech elements into words and sentences, and associate those with meaning, which depends on our language exposure and ability. It is not surprising, then, that the hearing test, measuring the detection of tones in quiet, is not a good predictor of speech-in-noise ability, or the benefit from hearing devices. Two patients with the same hearing loss who are given the same hearing aids with the same improvement in their hearing tests can have strikingly different speech in noise ability after: so that one is happy with the result and the other throws the hearing aid away.

In this programme we will develop tests of the ability of subjects to group together and retain sounds that have a complex structure, like speech, and separate these from a noisy background. Unlike speech, the sounds are not associated with meaning, and their detection can be measured precisely in patients that speak different languages and have different language skills. These tests allow us to measure an important process that is more closely related to speech-in-noise analysis than simple hearing tests. The first tests we have used predict how well people hear speech in noise independently of simple hearing tests. We aim to develop further tests for the hearing clinic to predict how well people will manage in the real world: real world listening tests.

Although the tests we will develop assess what might be considered a simple process based on grouping and separation of sound elements, the analysis of these requires extensive processing in the brain, which we want to understand better. This analysis requires the auditory part of the brain, but also parts of the brain that are not usually considered parts of the auditory brain, including key structures usually associated with memory. Perhaps this is not surprising: sound stimuli, unlike static visual stimuli, evolve over time, and the separation of mixtures of sound requires us to remember or hold in mind the elements that form the foreground and background. In this work, we will measure brain activity that allows us to separate mixtures of sounds like speech in noise. We will use functional magnetic resonance imaging carried out on normal listeners when they carry out these tasks as an indirect measure of brain activity. We can also measure brain activity directly, in patients with epilepsy who have electrodes placed in the brain for a few days in order to work out where the epilepsy starts. As well as the localising their epilepsy, we can measure brain activity during listening tasks.

The outcome of these brain experiments will be the definition of brain systems for separating sounds like speech in noise. This will, firstly, provide other measures (in addition to the new listening tests) tests) of the success of interventions like hearing aids, cochlear implants, or hearing training. Secondly, the work will suggest possible interventions that might help patients to understand speech in noise. Although it may seem far fetched to try to improve speech-in-noise detection by interfering with the brain, there is already evidence that minimally invasive electrical stimulation can do this, and the work could identify brain targets for drug treatments.

Behavioural measures of pre-language auditory cognition will be developed that can account for the ability of subjects to hear speech in noise (SiN). This follows pilot work establishing that prototype non-speech auditory figure-ground (AFG) and auditory working memory (AFM) tests predict SiN independently of the audiogram. New tests will use stimuli that are closer to the spectrotemporal structure of speech.

fMRI at 3T will seek brain bases for the correlation between the non-speech behavioural measures and SiN. The work will test for common changes in BOLD activity in auditory regions as a function of task difficulty for AFG and SiN and for AWM and SiN. The work will also seek evidence for a common system for AFG and SiN and for AWM and SiN using modelling to identify common effects of the tasks on effective connectivity within and between auditory areas. fMRI at 3T on the AWM system will test the involvement of the hippocampus in AWM and in active tracking tasks. If the 3T experiments are successful, further work at 7T will allow more sensitive measurement of BOLD changes within subareas during the same tasks and more precise definition of time series to allow modelling of effective connectivity within and between auditory areas.

Neurophysiological recordings from neurosurgical patients will measure local field potentials (LFPs) to seek brain bases for the correlation between non-speech behavioural measures and SiN. Two groups of case studies with ten subjects each will be carried out to assess brain activity in auditory cortex during AFG and SiN tasks, and to assess activity in the network including auditory cortex, inferior frontal cortex and hippocampus during auditory working memory. Single unit recordings will be carried out from hippocampus and parahippocampus during AWM and auditory tracking tasks to seek human units that allow auditory computation.

The work will benefit
- Those who develop clinically significant hearing loss during their lifetime (approximately half the population)
- Hearing professionals who assess them
- Policy makers
- Scientists and engineers interested in developing brain interventions to help real-life listening
- The UK economy

Mechanisms for benefit
- For patients, clinical tests developed during the lifetime of the project will better predict the benefit that patients might obtain from hearing restoration devices (especially common hearing aids), which will encourage their use in good prognosis, or suggest additional hearing rehabilitation otherwise
- Hearing professionals will be provided with objective measures to provide such prediction, to monitor responses, and to design rehabilitation strategies
- For policy makers, the measures will provide better data on the effects of hearing loss on the population using measures that can be applied to large numbers of people that are closer to real life listening than standard hearing tests, but easier to interpret than speech-based tests or questionnaires
- Also for policy makers, the work has the potential to explain how hearing loss, due to ear damage, contributes to degeneration in the brain, in dementia, strengthening further the public health argument for doing as much as possible to maintain hearing in middle age populations to prevent dementia
- For scientists and engineers, the work will identify brain targets for realistic interventions to help speech in noise like minimally invasive electrical stimulation
- The economy will benefit from measures to better predict and monitor the effects of hearing loss as a result of which more patients might actually use hearing restoration devices that are prescribed: hearing loss is estimated to cost the UK economy £30 billion per annum and a proportion of this is related to devices that are not used

Ways of maximising impact
- Public engagement to explain the research based on talks and website
- Making hearing professionals aware of new tests via clinical teaching and events carried out by group and via website
- Making scientists and engineers aware of brain study results via publications, scientific meetings and website
- Development of screening tools based on the tests developed that can be used in the community by smartphone