Many natural sounds are composed of a large number of stochastic events, e.g. the sound of wind, rain or insect swarms, called acoustic textures. Humans can easily recognize these sounds, despite substantial variability between different examples from the same type. An acoustic texture is characterized by its statistical composition, which has been shown to be sufficient for recognition. However, the neural representation of these statistics remains unaddressed. We recently investigated human performance of detecting changes in acoustic textures, and estimated cortical locations that represent the sound and the decision process. Here, we will address the low-level, neuronal representation in these locations directly by recording from populations of optogenetically identified neurons simultaneously in the mouse. The animals listen to naturalistic acoustic textures which change at a random time, and indicate their percept by licking to receive a reward. Simultaneously, we will monitor the neuronal responses in primary and secondary auditory cortex and parietal areas. The recordings will provide insight into (1) the representation of statistical stimuli in neuronal responses, (2) the dynamic estimation process of textures and (3) the transformations in stimulus representation up to the point of decision making. Passive experiments will be performed as well to demonstrate the contribution of active processing. In summary, the proposed project will substantially improve our understanding of auditory processing in complex, natural acoustic environments, and more generally the representation of probabilistic quantities in the brain. Applications include the design of speech recognition systems and hearing aids which work robustly in everyday environments.