What does the olfactory cortex know about behavioral context?

NOEL FEDERMAN; ANTONIA MARIN BURGIN; MACARENA AMIGO DURAN; SEBASTIAN ROMANO

Congreso; Bernstein Conference 2020; 2020

Bernstein

Sensory representations in primary cortices have been classically considered as patterns of neuronal activity that invariably encode physical characteristics of the environment. Increasing evidence is now showing that as animals learn the association between a sensory stimulus and its behavioral relevance, stimulus representation in sensory cortical areas can change. Moreover, internal state as well as animal movements related to a particular task have been shown to modulate cortical activity. Our aim is to study the contribution of different sensory and non-sensory variables related to a task in the activity of the primary olfactory cortex of mice. To do this, we developed a spatial context-olfactory task in which mice learn that an odor is rewarded when presented in a specific spatial context. Animals are head-fixed, but walk freely through a virtual corridor. We first analyzed different aspects of mice behavior during task learning. We found that animals learn to discriminate odors and reward position earlier than to discriminate visual contexts, and that task-related variables as locomotion speed and inhalation rate are adjusted along learning. We also measured the activity of neurons in the piriform cortex (PC) by extracellular recordings obtained with a multielectrode array, at first session of learning and at expert level. We found that olfactory cortex neurons respond not only to odor but also to visual context and reward, indicating that PC encodes information about relevant aspects of the task. We further dissected the contribution of different sensory and non-sensory variables to the modulation of PC activity at the level of each trial using a statistical approach based on Generalized Linear Models (GLMs). Our results show that, after animals have learned the context-odor-reward association, animal position in the virtual environment modulates PC responses to the extent that the spatial context could be successfully inferred from spiking activity using optimal decoders based on the fitted GLM models. Furthermore, we found that olfactory and non-olfactory variables, such as odor, inhalation rate, visual context, position in the corridor, locomotion speed, licking and reward, modulate PC activity in different ways and that learning induced a reorganization of the contribution of visual context and reward modulation. This suggests that the PC may use information from other brain areas to adapt odor processing depending on experience and behavior.