Damaged Ears at the Cocktail Party - Investigating How Cochlear Damage Impairs the Neural Representation of Speech in Background Noise

The ability to understand speech is one of the most important functions of the human auditory system, but this ability is considerably degraded in people with hearing deficits. Estimates from the World Health Organization indicate that 360 million people worldwide exhibit some type of hearing loss. Hearing aids have provided some benefit for patients affected with hearing loss, nevertheless, this benefit has been, to a large extent, restricted to making sounds more audible via amplification with little, if any, benefit to increase the performance to separate speech from background noise.
Recent data suggest that a range of hearing problems, including speech-in-noise deficits, may arise as a result of the loss of auditory nerve fibres (ANFs), i.e., the bundle of fibres that carry information from the inner ear to the brain. This represents a type of peripheral damage that can arise even in supposedly normal-hearing subjects (i.e. those with no evidence of reduced sensitivity in the audiogram). Moreover, it is likely that phenomena, such as adaptive coding, a process in which neurons adapt their rate of discharge of action potentials (a measure of responsiveness) to better represent the most commonly occurring sound intensities in the acoustic environment thus allowing the average background intensity to be averaged out, also contribute towards the processing of signals in the presence of a noisy background. We hypothesize that A) the loss of ANFs, in particular, those involved in the processing high-intensity sounds, is involved in a decrease ability to understand speech-in-noise due to a degraded representation of the acoustic signal and B) that hearing loss will decrease the adaptive coding capability of the auditory system, which may be critical for the discrimination of speech from background noise.
To test these hypotheses, we will investigate mechanisms used by the auditory system to process speech signals embedded in noise. For this purpose, we will use an animal model and manipulate its auditory system at the level of the periphery. Through the administration of drugs, the use of earplugs or by exposure to high intensity noise, we can induce different types of hearing loss by targeting different populations of cells in a highly controlled manner that can be assessed using standard methods. Then, to investigate the effects of these manipulations we will use sophisticated in vivo electrophysiological techniques to record the activity of individual neurons in the inferior colliculus (IC) of the midbrain, a major relay nucleus on along the ascending auditory pathway. This technique allows us to record from neurons "extracellularly", i.e., the activity of neurons in the close vicinity of the recording electrode. We will characterize the responses of auditory neurons to different types of stimuli, including pure tones, speech tokens and broadband noise. Subsequently, we will use identical acoustic stimuli and intracellular recording techniques to record the activity elicited from within individual neurons. This technique has the benefit of allowing us to record supra-threshold (action potentials) and sub-threshold events from individual neurons. The latter reflect inputs from many other neurons and can either increase (excitatory) or decrease (inhibitory) the likelihood of more action potentials been fired by a neuron. The analysis of these electrophysiological measurements will allow us to understand a) how well snippets of speech can be discriminated based on the responses of neurons in the midbrain, b) the capability of neurons to exhibit adaptive coding and how this relates to speech discriminability, and c) the effect hearing loss on these measures. This will help us to uncover the peripheral and central factors that contribute to speech-in-noise performance. The results we obtain from this research will be important for the development new strategies to improve the performance of hearing aids and cochlear implants.

The ability to understand speech in complex noisy environments is considerably degraded for those with hearing impairment. Hearing thresholds alone are insufficient to explain the range of speech in noise performance seen in patients, and amplification through hearing aids only provides limited benefit. Additional factors at the peripheral as well as central level might therefore play a role. Recent data suggest that a range of hearing problems, including speech-in-noise deficits, may arise as a result of deafferentation of auditory nerve fibres (ANFs). Moreover, it is likely that neural adaptive processes, such as adaptive coding, contribute to the processing of signals in a noisy background. We hypothesize that A) deafferentation of high-threshold ANFs decreases understanding of speech in noise due to a degraded representation of the signal and B) that hearing loss decreases the capability for adaptive coding, resulting in a reduced capacity of neurons to average out mean background intensities which may be critical to separate speech from noise. To test these hypotheses, we will employ an animal model to study the neural representation of speech in noise, how it is affected by hearing loss, and the role of adaptation to sound level statistics in the central auditory system. Specific types of hearing loss will be induced using noise exposure, ototoxic drugs, or earplugs. We will then investigate the effects of these manipulations using single unit as well as intracellular recordings in the inferior colliculus. We will analyze a) how well speech tokens can be discriminated based on the neuronal activity, b) the adaptive coding capability of the neurons and how it is related to speech disciminability, and c) the influence of hearing loss on these measures. This will provide powerful insight about the peripheral and central factors contributing to speech in noise performance.

Hearing loss is the most common sensory disorder, it is estimated that more than 300 million people world-wide suffer from disabling hearing loss, and problems understanding speech in the presence of background noise are one of the main complications brought along by the condition. However, the severity of speech-in-noise problems varies considerably across patients, and hearing thresholds alone cannot account for this variability. Unfortunately, hearing aids provide only very limited benefit, in challenging listening situations they might make sounds audible but not increase intelligibility. Poor performance in background noise is one of the main reasons for dissatisfaction with hearing aids, and these expensive instruments then often end up in the drawer due to perceived lack of benefit.

Poor understanding of the factors that determine speech-in-noise performance is probably one of the main obstacles in the way of developing better rehabilitation strategies. The current research therefore systematically investigates possible reasons for speech in noise problems in an animal model, investigating factors in the peripheral as well as the central auditory system. This will lead to a better understanding of the origin of the problem, and pave the way towards better rehabilitation strategies, comprising advanced assessment of hearing loss, new hearing aid fitting strategies, and the development of new algorithms for hearing aids and cochlear implants. Importantly, drug-based therapeutic interventions are also a distinct possibility - the Ear Institute is already working with commercial partners (including in Phase 1 clinical trial) with pharmaceuticals directed at central hearing dysfunction.

The main beneficiaries of this research will therefore be hearing loss patients, who will get long-term benefit from improved treatment, with the potential to significantly improve their quality of life. There are thus several million potential beneficiaries in the UK alone. Other beneficiaries will be audiologists, hearing therapists and ENT doctors, who will benefit from a better understanding of how speech in noise problems are related to cochlear damage and deficits in the central auditory system, which will provide guidance for the assessment of hearing loss as well as the choice of rehabilitation strategies. Moreover, R&D departments of hearing aid and cochlear implant companies will receive vital impulses for the development of devices that offer increased benefits in challenging listening situations. Our research will also be relevant for a variety of academic beneficiaries, including hearing researchers, auditory and systems neuroscientists, biomedical engineers, and researchers working on speech-recognition algorithms. Finally, our results could also have an impact on NHS policies, since they could influence guidelines for the rehabilitation of hearing loss and lead to more targeted interventions that might contribute to reducing healthcare costs.

The delivery of some impact aspects can already be started during the lifetime of the project, like for example communicating our results and the implications for assessment and management of hearing loss to clinicians through conference presentations and CPD courses. Dissemination of results to hearing loss patients, for example through the charity Action on Hearing Loss, can also be begun towards the end of the project. Development of new treatment strategies and devices could take place within a 5-10 year timeframe.