Microcircuit mechanisms behind the distributed cortical contributions to working memory

Project Summary / Abstract Working memory is the ability to hold and manipulate information in short-term memory in the absence of sensory input. Recent investigation into the neural underpinnings of this key cognitive ability has highlighted the existence of distributed working-memory representations across the cortex of macaques and humans. However, we still do not understand how microcircuits in different cortical areas potentially support their distinct contributions to working memory. To answer these questions, my laboratory has developed a new working-memory paradigm for mice navigating in virtual reality, in which short-term memory is temporally disentangled from other task computations, and its duration manipulated systematically. This will allow for the integration of the unique genetic and optical toolkits available for mice. Our preliminary data from mesoscale calcium imaging of animals performing this new task show that the mouse dorsal cortex also contains distributed working-memory representations. Further, these distributed representations map onto a gradient of spontaneous activity timescales that increase from sensory to frontal areas. This is such that the primary visual cortex (V1) is active exclusively during short memory delays, whereas frontal areas such as the premotor cortex (M2) are preferentially recruited when sensory information needs to be remembered over seconds. Here, we will ask how genetically identified cell types in V1 and M2 distinctly interact to generate different activity timescales and patterns of recruitment during working memory. In particular, we will test the hypothesis that differences in the amount of recurrent excitation and the ratio between inhibition mediated by somatostatin- and parvalbumin- positive interneurons explain activity patterns during both spontaneous and working-memory behaviors. We will use simultaneous two-photon Ca2+ imaging and optogenetic stimulation of single neurons to compare how V1 and M2 differentially respond to focal excitatory input. Moreover, we will silence different subtypes of inhibitory interneurons in each area to understand their role in modulating excitatory activity timescales, measured with two-photon microscopy or electrophysiology. Finally, to understand how excitatory neurons interact with different inhibitory-neuron subtypes to result in distinct patterns of task engagement of V1 and M2, we will simultaneously image from pairs of neuronal types using two-photon microscopy as mice perform our new working-memory paradigm. Beyond its relevance for basic neuroscience, our work will have translational implications: alterations in both cortical intrinsic timescale hierarchies and working memory have been reported in normal aging, autism, and schizophrenia.