A Neurocomputational Model of Multisensory Illusions in Subjects with Mild Hearing Loss

This thesis is framed within the study of multisensory integration, with particular attention to how the nervous system combines information from different sensory channels to build a coherent representation of the environment. Under normal conditions, this process improves perceptual accuracy and reduces uncertainty, but it can also give rise to illusory phenomena when sensory signals are temporally mismatched or otherwise incongruent. Among these, the sound induced flash illusion (SIFI) offers a particularly useful paradigm for investigating the mechanisms underlying audiovisual integration and causal inference. Within this context, the present work examines the differences between individuals with mild hearing loss (MHL) who regularly use hearing aids (HU) and non-users (NU), with the goal of understanding how prolonged experience with auditory amplification may influence perceptual behavior and cross-modal integration processes. Based on available behavioral data from the literature, this thesis develops and analyzes a neurocomputational model designed to reproduce the main patterns observed in the two groups and to interpret them in terms of multisensory dynamics and reliability-based weighting of sensory modalities. The overarching aim is to connect experimental evidence with theoretical modeling in order to provide a mechanistic and physiologically plausible account of the differences between HU and NU. From this perspective, the model does not merely describe the observed behavior, but also seeks to offer an interpretive framework for the functional changes that may accompany prolonged hearing aid use.