Title: Sensory representation and learning in an olfactory cocktail party task
Abstract:
The olfactory world presents us with a countless number of smells, yet we only have a limited number of receptor input channels to present them. These smells often occur simultaneously, making it even more complicated to detect or separate odor objects. How does the brain solve this sensory representation question? We train mice to perform an olfactory cocktail party task, which captures the essence of olfaction-form figure-ground separation. While previous work has recorded the activity of olfactory sensory neurons in mice performing this task, we move one step further into the cortex and record the activity of piriform neurons in naive, learning and expert animals. We are starting to do some preliminary analysis on this dataset. Besides, we also treat the learning of this mixture task as a model of learning, and use behavioral approaches to ask what is the mechanism that the animal brain uses to learn the categorization rule.