Document Actions

You are here: Home / Bernstein Conference / Past Conferences / 2016 / Satellite Workshops / Natural Scene Statistics and Sensory Representations (Mlynarski)

Natural Scene Statistics and Sensory Representations (Mlynarski)


It is a fundamental principle of evolution that organisms adapt to their environments. According to this rule, sensory systems are shaped by regularities of the organism's natural sensory niche. To date, numerous studies have indicated close connections between statistics of natural stimuli and sensory representations employed by the brain. Growing evidence suggests also that neurons respond differently to natural signals than to artificial stimuli such as pure tones or sinusoidal gratings. Since sensory systems evolved and developed to process natural scenes, the understanding of their mutual relationship is necessary to fully elucidate the mechanisms of neural coding.

The main purpose of this workshop is to focus specifically on intermediate level representations of natural stimuli - both artificial and biological. While a lot of work has been devoted to early-level representations such as Gabor filters, not much is known about stimulus structures of intermediate complexity and how they are encoded by the brain. Our primary aim is to address this gap and discuss recent advancements in physiology, perception and machine learning of mid-level codes. We will primarily consider representations specific for vision and audition, but we also will discuss possible common mechanisms shared between different sensory modalities.