Neural computation in odour space
Dr. Michael Schmuker, School of Engineering and Informatics, University of Sussex, UK
Our sense of smell enables us to explore the world of chemical information. Yet, our knowledge on the structure of chemical stimulus space still lacks far behind other modalities like vision or hearing. This lack of knowledge currently presents a major roadblock for understanding how the brain efficiently encodes chemical information. Moreover, a better understanding of odour space, and how it is processed in the brain, may also enable bio-inspired design of efficient technical solutions for chemical sensing.
In this presentation, I will give an overview on our research on how the olfactory systems of insects and vertebrates encode and transform chemical information on its way from the primary sensors to higher brain areas. These investigations inspired us to implement the key concepts of olfactory processing on a neuromorphic hardware system that uses spiking neuronal networks to perform pattern recognition in high-dimensional feature spaces. I will also present our recent findings on how to extract information about source distance from the fine-structure of gas plumes.