In this episode, we delve into the fascinating world of auditory neuroscience with Dr. Elena Dellaferrera. We explore how our auditory system separates individual sources from mixed sounds - a challenge known as blind source decomposition.

In nature, auditory signals usually originate from multiple sound sources occurring simultaneously. Recognizing these individual sources is a formidable task for the human auditory system, which it handles elegantly by identifying repeating patterns within the acoustic input. However, recreating this behavior through computational models has been a largely unexplored territory.

Dr. Dellaferrera introduces us to her innovative, biologically inspired computational model designed to perform blind source separation on mixed acoustic stimuli sequences. The proposed model, grounded in a somatodendritic neuron model and a Hebbian-like learning rule, was designed to detect recurring spatio-temporal patterns in synaptic inputs.

This computational model effectively segregates sources, echoing the characteristics of human performance across different experimental settings using synthesized sounds with naturalistic properties. Dr. Dellaferrera's work further extends this study to unexplored territories, including natural sounds and images.

Join us as we delve into the implications of this exciting research, its potential to enrich our understanding of auditory neuroscience, and how it can inform predictions in yet-to-be-tested experimental settings.

Keywords: Dr. Elena Dellaferrera, Blind Source Decomposition, Auditory Neuroscience, Computational Models, Sound Source Separation, Somatodendritic Neuron Model, Hebbian Learning Rule, Segregation Capabilities.

Modeling the Repetition-Based Recovering of Acoustic and Visual Sources With Dendritic Neurons https://doi.org/10.3389/fnins.2022.855753