Network Effects Of Reduced Audibility (NEORA) and its Significance for Cognitive Decline

Abstract

Though hearing loss has been found to increase a person’s risk of Alzheimer’s Disease and other dementias, the nature of this relationship remains to be established, ultimately centring on questions of causality. Isolating the role hearing loss can play in cognitive decline—separate to the multiplicity of other, often interrelated biological and lifestyle factors—presents a recognised challenge to correlational research methods. In this thesis I introduce a computer model for the Network Effects Of Reduced Audibility (NEORA) that quantifies the impact of information degradation in the cognitive task of speech understanding. My novel approach combines information theory with network theory to build a causal model of speech understanding based on the underlying relationships between phonemes, words and their statistical distribution within a corpus of spoken English. Auditory information is then defined in terms of unique and unambiguous acoustic signals that enable a listener to reliably choose between possible alternatives (phonemes or words) in order to match their own mental representations to those intended by the speaker’s message. When audibility is reduced, information is degraded through phoneme removal, leading to increased ambiguity (listener uncertainty), quantifiable using the new Speech Understanding Index (SUI). I demonstrate a causal relationship between reduced audibility (measurable by an audiogram) and information degradation that results in the need to recruit additional sources of information, e.g. vision and memory. As an example, I determine the hearing levels at which lip-reading becomes more critical for speech understanding than hearing. The thesis concludes by discussing the significance of quantifiable information degradation as a basis for understanding the phenomenon of dedifferentiation in cognitive decline. I predict that reduced audibility will increase and shift metabolic energy demands towards brain regions most associated with integration of multiple sources of information and the resolution of ambiguity, suggesting that a search for a causal link should begin there. Thus NEORA and the SUI provide researchers with a quantitative framework for modelling and measuring causality, crucial for unravelling the relationship of hearing loss to cognitive decline.