Christopher Harvey - US grants

? DESCRIPTION (provided by applicant): Working memory, the temporary storage of information for future manipulation to guide actions, is an essential component of nearly all cognitive processes. Deficits in working memory contribute to a variety of mental health disorders, including schizophrenia, autism, bipolar disorder, and attention-deficit hyperactivity disorder. Working memory is thought to be an emergent property of neurons acting in groups to form microcircuits. However, due in large part to technical limitations, working memory has been studied nearly exclusively at the scales of entire brain regions or individual neurons, both of which fail to reveal interactions in neuronal populations. A major challenge in understanding the neural mechanisms for working memory is to develop tools to measure, analyze, and model working memory at the scale of the neuronal microcircuit. In this application, we present approaches to solve this challenge using two-photon imaging of activity in neuronal populations in mice performing working memory tasks in virtual reality. We will combine our large imaging data sets with theoretical modeling approaches to develop new computational models of how working memory is generated in microcircuits. First, we will examine how working memory representations are encoded in neuronal populations by examining the stability of representations over long timescales, using chronic two- photon imaging of the same neurons over weeks during working memory behaviors. Second, we will perform imaging experiments during novel working memory tasks to test competing, long-standing theoretical models of working memory, including models based on attractor dynamics and sequence dynamics. Third, we will develop a new theoretical modeling framework in conjunction with rapid experiment-model iterations to generate and test new microcircuit-scale hypotheses for the neural mechanisms underlying working memory. Together the proposed work is expected to establish new approaches to measure, analyze, and model microcircuit function during working memory in the mouse, leading toward mechanistic studies of how working memory and its underlying microcircuits are disrupted in mental illness.

? DESCRIPTION (provided by applicant): Sensorimotor processing during spatial navigation requires the communication between a large number of brain areas. The posterior parietal cortex (PPC) is a crucial interface for arranging this communication. Leading models, based on extensive work in primates and more recently in rodents, hypothesize that the PPC links sensory and cognitive signals, such as those that guide navigation, with the decisions and plans that drive upcoming locomotor actions. The PPC is therefore predicted to receive sensory signals as inputs and to transmit signals regarding navigation decisions and plans. Although all models of PPC function rely on the interaction between the PPC and multiple other regions, it is currently poorly understood what input signals the PPC receives and how the PPC routes output information to target regions, in large part due to technical limitations. Here we will develop and implement new approaches to measure directly the signals contained in the PPC's input and output channels. Our approach will be to use optical imaging and anatomical tracing technologies in the mouse in combination with behavioral tasks in virtual reality environments. In a first aim, we will examine how the PPC routes information related to sensory and motor events during navigation decision tasks. We will test if the PPC selectively routes different types of information to different target regions or if the PPC transmits information generally to establish widely distributed network. In a second aim, we will measure the input signals the PPC receives from the auditory cortex. We will test if these signals selectively encode specific features of the sensory world, such as spatial information, and how the input signals to the PPC are modulated depending on behavioral relevance and behavioral state. Together this work will advance our understanding of information transfer in a multi-region network for sensorimotor processing, which is essential for nearly all complex behavioral tasks.

Project Summary During perceptual decision-making, populations of neurons, arranged in highly interconnected microcircuits, work together to encode sensory stimuli and to transform sensory perception into appropriate behavioral choices. A fundamental gap in our knowledge about perceptual decision-making is understanding how the connectivity in cortical microcircuits shapes dynamics and information codes in populations of neurons. This gap has arisen because anatomical connectivity and activity have generally been studied separately, and because a computational framework to understand structure-function relationships in cortical microcircuits is missing. Here, we will assemble a team of researchers with complementary skills to tackle this problem. We will combine approaches to study population coding and dynamics using two-photon calcium imaging during a novel and complex decision task for mice, with measurements of connectivity in the imaged neurons using electron microscopy (EM)-based connectomics. Furthermore, we will use our activity and connectivity data to develop a data-driven model to explore structure-function relationships across cortical microcircuits. We will apply our new approach to investigate how population codes, microcircuit connectivity, and structure- function relationships differ across cortex to perform distinct computational tasks during perceptual decision- making. Although it is well established that sensory and association cortices perform different functions, little is known about the mechanisms underlying these different roles, including distinctions in microcircuit connectivity and population coding schemes. In a first aim, we will compare population codes and microcircuit connectivity for sensory stimuli and behavioral choices in visual cortex (V1; sensory cortex) and posterior parietal cortex (PPC; association cortex). We will use computational tools to examine how distinct coding schemes provide functional benefits. We will use EM connectomics in V1 and PPC for neurons imaged during a perceptual decision task to probe structure-function relationships for stimulus and choice codes. We will develop a data- driven recurrent neural network model to relate connectivity and population activity. In a second aim, we will investigate how neuronal populations transform sensory information into behavioral choices using microcircuit connectivity. We will develop a new statistical concept ? intersection information ? to identify activity patterns in V1 and PPC that carry sensory information that informs behavioral choices. Using EM connectomics, we will reconstruct the microcircuit connectivity between cells to test hypotheses about sensory-to-choice information flow. Our work will be some of the first to compare population coding and microcircuit connectivity across cortical regions and to explore structure-function relationships for perceptual decision-making.

Summary The cognitive functions of the brain allow higher animals to interact with the world in complex and adaptive manners. For example, animals use past experiences to develop internal models of rules and contexts that shape subsequent decisions. These cognitive processes are disrupted in many devastating mental health disorders. Traditional experiments in cognitive neuroscience have identified correlations between cognitive processes and behavioral outputs or neural events. However, it has been challenging to uncover the causal mechanisms by which cognitive processes emerge from the basic building blocks of neurobiological computations: molecules, cells, synapses, and circuits. The reason is, in large part, because it has been difficult to combine emerging mechanistic approaches in neuroscience with paradigms to study cognitive processes. Here we aim to overcome these challenges by developing a research program to identify causal links between cognitive processes and the structure ? cell types and connectivity ? and function ? spatiotemporal activity patterns in neural populations ? of neural circuits. We recently devised a virtual reality system for mice and developed methods to train mice to perform complex, cognitive tasks as they navigate through virtual environments. Further, we developed neurophysiological and computational tools that have identified correlates of cognitive processes in the activity patterns in population of neurons. We will use this foundation to establish a research program for mechanistic cognitive neuroscience. First, we will develop an atlas of cell types in cognitive brain regions, use viral tools to label these cell types, and then study the functional roles of these cell types during flexible decision-making tasks. Second, we will establish methods based on single-neuron optogenetics to reveal connectivity between identified cells during cognitive behaviors. Third, we will use calcium imaging to ?read? patterns of neural activity during cognitive tasks and will then ?write? and ?erase? these patterns using patterned optogenetics to test sufficiency and necessity. We will study these structural and functional properties in local populations of neurons (microcircuits) and across brain areas (mesocircuits). This work will expand the emerging field of mechanistic cognitive neuroscience and develop a new research program toward the goal of defining cognition and mental health in terms of core biological components and mechanisms.