Hearing in a social context: Understanding predictive mechanisms in communicative interaction

Conversation is fundamental for developing social attachments, and if the ability to converse with others breaks down, it can be devastating. Not only does conversation with others improve psychological wellbeing, it also buffers people against the ill-effects of stress and other health issues. People with hearing impairment report social isolation and loneliness as one of its most disabling outcomes, yet we know very little about how to support conversation for people that struggle. To date, the majority of research examining speaking and listening behaviour has tested individuals in isolation, using experimental methods that do not require them to actually engage with someone else. But it is only by investigating speaking and listening in a social context that we can address the cognitive processes specific to conversation.

To interact smoothly, people must make mental predictions about each other's contributions. Conversations would be much less fluent if people waited for the person they were talking to to finish before they began preparing their response. Yet people with hearing impairment show reduced use of prediction during speech listening, even though they could particularly benefit from such help given their poor auditory input.

The objective of this project is therefore to understand how we make predictions during interaction, in order to identify ways of making conversation easier for people with hearing impairment. Combined behavioural and neuroscientific experiments will explore how people make predictions during interaction at both the mental level and the brain level, and I will draw this work together to generate a model of prediction in interaction. The way that predictions differ with hearing impairment will also be explored: do people with hearing impairment make predictions too late to be useful, are they less confident in their predictions, or are their predictions simply inaccurate? Following training with a hearing device manufacturer, subsequent work will identify how the findings could be implemented in hearing technology to support people with hearing impairment have conversations successfully. This work could transform the state-of-the-art in hearing devices, and therefore has the potential to benefit the psychological wellbeing of millions.

I have identified two main beneficiaries for my 7-year research programme: (1) the hearing loss community, and (2) industries focused on interaction.

(1) Approximately one in six people in the UK have some form of hearing loss, which leads to loneliness and depression. Hearing devices offering improved comprehensibility for patients reduce such negative outcomes, and save businesses money in lost days and staff turnover. This research could therefore be used by hearing device manufacturers to motivate critical changes in the handling of the audio signal to support individuals with hearing loss. The patients themselves, their families, and the carers of individuals with hearing loss would benefit from such developments as it would improve engagement quality, and make it easier to share important information. While these stakeholders could benefit from the basic science findings of stage 1, I primarily aim to impact this group in stage 2. As the potential for quality of life benefit and increased wellbeing for a large population is significant, I will convey my research more widely by engaging with parliamentary interest groups (e.g. the All-Party Parliamentary Group on Deafness), and charities such as Action on Hearing Loss.

(2) This work could impact other industries focused on interaction, inspiring the development of new technology to improve the success and naturalness of their products. For example, teleconferencing systems can be problematic in noisy rooms. The hearing device algorithms developed in stage 2 could be incorporated into such systems to amplify relevant auditory information and attenuate irrelevant auditory information, improving user experience. In addition, my research could also be valuable to AI firms who work on predicting pathways through communication, as they could use my model of prediction (and relevant anonymised data) to train new algorithms.

Findings will be shared with these audiences through industry conferences/meetings, clinical workshops, engagement with parliament and charities, and press releases.