Computations of transcriptomic neuron types in cortex

The cerebral cortex contains a rich diversity of neuronal types, particularly among inhibitory cells. Transcriptomic methods, which measure the expression of many genes simultaneously, have identified nearly 100 fine subtypes of cortical neuron. Understanding how these subtypes underlie cortical computation requires measuring the activity of large cortical populations in vivo, identifying each cell's transcriptomic subtype.

We have developed a method to identify the transcriptomic subtypes of all cells in a population whose activity has been measured in vivo by 2-photon calcium imaging. We will use this together with tools such as 2-photon mesoscopy, cell-type-specific viruses and mathematical data analysis, to reveal for the first time how the cortex's cell types are coordinated during information processing. Our working hypothesis is that an animal's cognitive state controls the joint activity of inhibitory neurons, which in turn sculpt information coding in excitatory populations. We will test this hypothesis and characterize in detail of each cell type's role in the circuit.

By recording during spontaneous behavior, we will ask if inhibitory population activity is constrained to a manifold reflecting fixed interactions between inhibitory subtypes, and how this inhibitory activity correlates with ongoing behavior and sensory coding in excitatory neurons. By recording transcriptomically-identified populations while mice perform a behavioral task, we will reveal how inhibitory activity and excitatory coding vary according to cognitive context such as stimulus expectation and omission, engagement, impulsiveness, and reward. By combining cell-type-specific optogenetics with transcriptomically identified recordings, we will reveal the causal role of each inhibitory subclass and of neuromodulators.

Together, these experiments will provide foundational information on how information is processed by the diverse cell types that make up cortical circuits.