A. David Redish - US grants

[unreadable] DESCRIPTION (provided by applicant): In recent years, advances in technology have made simultaneous recording of neuronal ensembles possible, leading to many exciting developments in the understanding of how the brain performs neural computations. These breakthroughs have depended on the development of a number of techniques for extracting information from distributed representations. One of the key techniques has been the development of reconstruction methods, by which one attempts to calculate a behavioral or stimulus variable from neural firing patterns. While reconstruction is a powerful tool, it only provides a value and gives no indication of the quality of the representation itself. Because neuronal representations are highly distributed, cells could be firing in a manner that is generally consistent or generally inconsistent with the reconstructed value. There is no way, using current reconstruction techniques, to determine whether the representation is robust or not. We have developed a measurement (coherency) which provides just such a detection method. This measurement can detect dynamic events (such as the resolution of ambiguity). Experimental aim: To determine whether the coherency of hippocampal ensembles differs between tasks. Computational aim: To improve the coherency measurement by increasing its temporal resolution and by deriving quantitative statistics so that it can be used to directly measure differences in representation. As more and more experiments are done with neural ensembles, novel methods are going to be needed to analyze and understand those ensembles. The coherency measurement enables the determination of how well those neural ensembles represent the values we believe they represent. Refining the coherency measurement will provide the neuroscience community with a useful tool for analysis of those neural ensembles. [unreadable] [unreadable]

[unreadable] DESCRIPTION (provided by applicant): The hippocampus is critically involved in flexible learning tasks, particularly in spatial navigation, reversal, and episodic-memory tasks and tasks that require planning and decision-making. In tasks with a spatial component, hippocampal cells show a strong spatial firing correlate (termed "place fields"). Because it is possible to record large neural ensembles from hippocampus, and because the set of place fields generally cover the entire environment, it is possible to reconstruct location from hippocampal neural ensembles. In spatial decision tasks, non-spatial information (episodic memory, decisions to be made, etc.) is projected onto the spatial domain. This means that the spatial tuning of place cells provides leverage with which to examine the dynamics of hippocampal representations on decision tasks. Using new reconstruction methods that enable the interpretation of representation of spatial location from neural ensembles at very fast (e.g. tens of ms) timescales, we have recently observed a reliable, repeatable phenomenon whenever rats paused at a decision point on a maze: During most behaviors, location reconstructed from the ensemble was an accurate representation of the rat's location within the maze. However, at decision points, the reconstructed location swept ahead of the rat, first down one potential path, and then down the other. After alternating back and forth for 500- 1000 ms, the representation returned to the location of the animal and the rat began running again. These non-local reconstructions reliably stretched forward of the animal rather than behind the animal. Our working hypothesis is that this phenomenon reflects a choice-consideration, path-, or goal-planning process. The objective of this proposal is to test this hypothesis and further our understanding of the mechanisms underlying these observations. Our plan of attack is to record neural ensembles in a cued-choice task and to determine the behavioral situations in which these phenomena occur. We will identify the location of the rat when these phenomena occur (e.g. does this only occur at choices?) and the locations which get reconstructed to (e.g. does it sweep all the way to the goal?). We will then characterize the hippocampal contribution to the phenomena through detailed comparison with afferent structures and a detailed analysis of hippocampal local field potentials, interneuron and pyramidal cell firing patterns, and the interaction between them. The work proposed here will increase our understanding of the role of the hippocampus in spatial navigation, episodic memory, and decision-making. PUBLIC HEALTH RELEVANCE Using novel analysis techniques applied to neural ensemble recordings, we have recently observed that hippocampal ensembles transiently encode the available choices when rats pause at decision-points. This objective of this proposal is to understand the mechanisms underlying this novel phenomenon, which may reflect a choice-consideration process. Understanding the mechanisms of the normal decision-making process will have implications for patients in which that process has broken down such as in anxiety disorders or other cognitive disorders such as Alzheimer's disease and Schizophrenia. [unreadable] [unreadable] [unreadable]

DESCRIPTION (provided by applicant): The decision between taking a small reward immediately or waiting for a larger reward later is a critical one faced by many creatures throughout their lives. This ability derives from an interaction between multiple decision-making systems and is related to self-control, and inversely related to addiction liability. Importantly, he balance between those systems changes over the lifespan, but the mechanism of that change is unknown. In rats, willingness to wait for a larger reward is usually measured by putting the two options into direct conflict, such as in the titrated delay task, in which the delay is titrated unil the two choices are equal. The problem with the titrated delay task as a means to study mechanism is that it does not generally titrate smoothly and cannot provide a complete discounting curve. We have developed a new spatial version of the titrated discounting task that titrates within a single session reliably and allows the direct measurement of discounting curve parameters across sessions. We propose to use this new discounting task to determine the changes in discounting rates across the lifespan. PUBLIC HEALTH RELEVANCE: The ability to wait for a larger reward is related to self-control, and inversely related to addiction liability. This ability derives from an interaction between multiple decision-making systems. Importantly, the balance between those systems changes over the lifespan, but the mechanism of that change is unknown. The goal of this project is to improve our understanding of these interacting decision-making systems by examining their changes across the lifespan.

DESCRIPTION (provided by applicant): Deliberation entails the serial examination and evaluation of outcomes. While there are detailed, mechanistic, and quantitative theories of non-deliberative decision-making, such theories of deliberative decision-making are still lacking. This is due to a lack of key experimental knowledge of the mechanisms of deliberation. Deliberative decision-making entails the sequential consideration of possibilities, requiring three steps in a repeated cycle: (1) prediction of the consequences of one's actions, (2) evaluation of those predicted consequences, and (3) selection of the best action. An important difficulty that has limited our ability to study deliberative decision-making is that the process of deliberation is covert, that is, the transient information being considered is not reflected in immediate behavior. However, new mathematical techniques now allow decoding of represented variables from neural ensembles at very fast timescales, enabling the observation of those transient, covert processes. The goal of this proposal is to track the covert prediction of reward outcomes as alternatives are evaluated. We have preliminary data that structures known to be involved in motivation and evaluation (ventral striatum, orbitofrontal cortex) show a transient activation of reward-related activity at certain deliberative decision-points. Combining newly available multi-structure recording techniques, newly developed tasks, newly improved neural ensemble analysis techniques, and computational modeling, we will examine the relationship between the representations of future possibilities in hippocampus and the covert reappearance of reward-related information in structures known to be involved in motivation and reward and decision-making. PUBLIC HEALTH RELEVANCE: The disease model of addiction suggests that addiction is fundamentally a dysfunction of decision-making. As our understanding of decision-making has improved, our understanding of how it can break down has also improved. The goal of this proposal is to improve our understanding of deliberative decision-making which will enable a better understanding of the potential dysfunction that can occur therein.

An extensive literature suggests a role for the prefrontal cortex and hippocampus in episodic memory, particularly in the re-construction of those memories; a growing literature is now suggesting a role for these structures in planning and imagination, particularly in the construction of future potential episodes, a process termed ``episodic future thinking''. These two processes have obvious similarities (they both entail construction of episodic representations) and obvious differences (memory is trying to reconstruct a real past, imagination is trying to hypothesize a potential future). However, it is not known how this neural circuit contributes to these phenomena, nor how the information processing differs between them. New mathematical techniques now allow decoding of represented variables from neural ensembles at very fast timescales, enabling the observation of transient, covert processes. The goal of this proposal is to use these new techniques to track changes in representation at cognitive (10 ms) timescales. Combining newly available multi-structure recording techniques, newly developed tasks, newly improved neural ensemble analysis techniques, and causal manipulations of the prelimbic cortex, we will examine the relationship between the representations of past events and future possibilities in hippocampus to determine the mechanism by which the hippocampus computes these representations.

7. Project Summary/Abstract Studies on synaptic plasticity underlying learning and memory have mainly focused on neuronal elements. However, astrocytes, classically considered as merely supportive cells, are emerging as key regulatory elements potentially involved in learning, memory, and neural information processing because they respond to neurotransmitters and modulate neuronal activity and synaptic function through the release of gliotransmitters. While important progress has been made to define the cellular mechanisms underlying astrocyte modulation of synaptic function in certain brain areas, such as hippocampus, retina and cortex, exactly what specific roles astrocytes play in the neural information processing system remains largely unknown. We have recently found that neuronal activity in the dorsal striatum elevates astrocyte Ca2+ levels, which, in turn, impact striatal neuronal activity and synaptic transmission. Yet, fundamental questions remain unknown, such as the astrocyte contribution to synaptic plasticity, to neural network information processing, and to the consequences on animal behavior. The present proposal aims to define the impact of striatal astrocyte activity on corticostriatal synaptic transmission and plasticity, striatal neural network activity and striatum-related behavioral task performance. We hypothesize that astrocyte activity controls synaptic function and plasticity, and influences neural network operation and animal behavior. To test this hypothesis we will combine in situ and in vivo approaches and state-of-the-art techniques, including optogenetics, pharmacogentics (?designer receptors exclusively activated by designer drugs?, DREADD), simultaneous two-photon microscopy Ca2+ imaging and multiple electrophysiological recordings, novel transgenic mice, in vivo electrophysiological recordings, and specific behavior studies. For this purpose, the proposal brings together two laboratories with complimentary expertise ? the Araque lab with expertise in astrocyte-neuron interactions in in vitro slices at cellular level and the Redish lab with expertise in striatal neuronal information processing and behavior at circuit and organism levels, that are already collaborating successfully, including co-advising a graduate student funded by a F30 NIH grant. The expected results will define the role of astrocytes in the striatal function and the consequent animal behavior, which will help to identify novel cellular mechanisms underlying brain function. Defining these roles of astrocytes on synaptic plasticity, network operation and animal behavior will reveal novel mechanisms involved in brain disorders occurring in certain brain diseases, such Parkinson´s and Huntington`s diseases, Obsessive Compulsive Disorder, Tourette's syndrome, and addiction, which may serve to identify novel cellular targets to develop future therapeutic strategies.

Project summary Current theories suggest that action-selection in the mammalian brain depends on an interaction between multiple, neurally-separable algorithms. The existence of multiple decision-systems opens up novel questions that do not exist within a unitary decision-maker: What happens when these systems select conflicting actions? How are those conflicts resolved? A number of disorders (OCD, eating disorders, drug addiction) and a number of RDOC-related dysfunctions (compulsivity, habits, and issues of cognitive and ?self-? control) have all been proposed to depend on resolutions of conflicts between these decision-systems. Recently developed human tasks have proved capable of putting these decision-systems into conflict for study. We have translated and validated a rodent version of this new human task. We will build on our established expertise in neural ensemble recording and computational analysis to examine how conflicts between these systems is resolved. Using DREADD manipulation and neural ensemble recording technologies, we propose to identify the mechanisms and computations that underlie conflict resolution between these decision-systems.

This proposal is for a new training grant under the Jointly Sponsored NIH Predoctoral Training Program in the Neurosciences (JSPTPN) at the University of Minnesota. Although a new application, this proposal is effectively a continuation of the University of Minnesota's highly successful NIGMS training program, ?Predoctoral Training of Neuroscientists? (T32-GM08471, 1993-2018) under the Systems and Integrative Biology Program. Trainees in this program are pursuing a PhD through the Graduate Program in Neuroscience at the University of Minnesota, an interdepartmental and interdisciplinary program that spans 30 departments. The program provides select trainees added value in the first two years of their graduate careers, providing its trainees with a broad foundation in neuroscience as well as the interdisciplinary skills needed to be successful within their neuroscientific careers. In particular, the program is designed to provide a broad understanding of the field as well as a deep understanding of research methodologies, experimental design, and quantitative reasoning. This program is built around a core of didactic coursework in neuroscience, exposure to research-related issues such as ensuring rigor and reproducibility and quantitative analyses, and the beginning of thesis-related in-depth research projects. Flexibility and time for in-depth development of collateral fields of knowledge are provided. Several unique educational opportunities are included, including the long-running Itasca summer laboratory, rotations, and specific classes in rigor, reproducibility, and quantitative reasoning. Time and attention are given to the student's professional development, including exposure to neuroscience at the national and international levels. A talented group of trainers have been assembled that reflects the diversity of research questions, areas of study, and techniques in neuroscience. Each trainer directs a productive research program and has demonstrated commitment to teaching and training. The trainers are united by their participation in the Graduate Program in Neuroscience and by their dedication to predoctoral training. An impressive array of scientific and institutional resources are available to the trainees, including substantial direct institutional support for this training grant. The graduates of this program will be trained to be independent researchers, capable of making contributions in academia, teaching, industry, government, and public service.

Project Summary New technologies have enabled amazing access to neural processes at multiple resolutions of time and space. However, the data analyses necessary to answer the scientific questions often depend on computational techniques that are unique to the experiment and may have to be modified (or even developed anew) for each specific experiment. These are not techniques that can be learned in a single class, but rather ways of thinking about problems that must be incorporated into each experiment and each analysis. Fundamentally, the data are being collected, but the field is not getting the full value of the collected data; students need additional training in order to successfully extract the complete information present in their experiments. Students need to understand both the complexities and limitations within experimental paradigms and also the complexities and limitations within computational analyses that can be applied to those paradigms. More importantly, if a student is going to develop his or her own analyses, the student needs a deep understanding of how to define and derive the appropriate control analyses. Ensuring the rigor of the experimental design and the subsequent analyses is particularly difficult for these types of data. We propose to build a comprehensive training program for both predoctoral graduate students and early-stage postdocs, to teach them how to integrate computational analyses and techniques to achieve scientific breakthroughs in neuroscience. This training will place these students in a very strong position for furthering their scientific careers. Furthermore, the training program will also help support, maintain, and improve the strong interdisciplinary community in computational, experimental, and clinical neuroscience that already exists within the University of Minnesota.

The overall goal of this project is the development and validation of parametric mathematical assays of human decision-making, based on an online information-foraging task (WebSurf). Decisional impairments in general are common in mental illnesses, but the exact pattern of deficits varies within and between diagnostic categories. Those deficits often involve multiple decision-making systems and the interactions between those systems. For instance, patients with obsessive-compulsive disorders rely overly on habitual/procedural action (leading to ritualizing behavior), but also show impairment in change-of-mind systems (inability to interrupt rituals) and deliberation (?analysis paralysis? in the face of uncertainty and a chance of negative outcomes). A major challenge in computational psychiatry is the need for tasks/paradigms that measure these multi- system dysfunctions, including interactions between systems. A further need is tasks that are viable for clinical settings, i.e. that are valid for repeated-measures use, sensitive to clinical-level impairment, and usable without highly trained experimenters present. We propose to address these needs with WebSurf, an information-foraging task developed by co-PIs MacDonald and Redish. These investigators and their colleagues have used WebSurf (and its rodent version, Restaurant Row) to demonstrate a common ?sunk costs? fallacy across rodents and humans, to identify the neural basis of regret, and to quantify differences in rule-based decision making in patients with eating disorders. Those studies have demonstrated WebSurf?s general utility as a cross-species paradigm and shown the richness of parametric descriptions that can be extracted from task behavior. They have also identified difficulties with the base version of the task, including needs for greater subject engagement and higher trial counts. As importantly, although Restaurant Row appears to elicit stable day-to-day behavior in mice, we do not yet know if the same is true for humans. We will close these gaps in task validation by assessing the performance of multiple variants using Amazon?s Mechanical Turk platform. Data from those variants, as well as ongoing data collection with the baseline task in our psychiatric clinics, will validate newer and more robust approaches to decision parameter estimation (Aim 1), grounded in Bayesian hierarchical modeling. They will demonstrate repeated- measures stability (Aim 2) and ability to describe variation between and within clinical populations (Aim 3). Executing these Aims will build on WebSurf?s success as a (reverse) translatable experimental paradigm, demonstrating a tool for clinical computational psychiatry. Our team?s broad experience includes computational science, experimental psychology and neuroscience, and clinical psychiatry, making us well-suited both to perform the Aims and apply the results in future psychiatric neuroscience studies.

PROJECT SUMMARY: COMPUTATIONAL CORE The purpose of the COMPUTATIONAL CORE is to provide services for infrastructure and conceptual support to integrate the four Projects that form our Center. To do this, the Core will ensure that the trial-by-trial behavioral data collected from the four Projects (from the computationally informative DPX and Bandit tasks) are analyzed in a uniform and compatible manner, so that findings across Projects can be compared and modeled. It will also supply overall statistical support to ensure that statistical analyses are done appropriately. The Core consists of a Service Aim and a Modeling Aim. In the Service Aim, we propose to apply causal discovery analyses, a recently developed toolkit of mathematical algorithms that can infer explanatory relationships between co-occurring data parameters. Causal discovery analyses serve as a powerful inferential toolbox that can be applied to all of the PROJECTS independently, facilitating the generation of new hypotheses within and across PROJECTS. In the Modeling Aim, the Core will provide conceptual and modeling support for the theoretical underpinning of state representation dysfunctions relevant to psychosis. As noted in the OVERALL RESEARCH STRATEGY, we take a central computational perspective to translate across species and methodologies, using theoretical constructs to bridge the gap between neural dysfunction and observable manifestations of that dysfunction. To this end, the Core will integrate two existing models: an Algorithmic-Level Model that translates attractor dynamics to behavior, and a Neurophysiology-Level Model that translates neuronal properties to attractor dynamics. Our goal is to examine how neuronal-scale effects, such as those seen in the non-human animal projects (PROJECTS 1 & 2) translate into observable behavioral and neurophysiological effects in healthy and clinical populations, such as those seen in the human projects (PROJECTS 3 & 4), by merging these two models into an integrated model crossing levels, that can provide mechanistic explanatory power for how neurophysiological effects produce attractor dynamics that lead to behavioral outcomes.

PROJECT SUMMARY: OVERALL To respond adaptively to the environment, the brain must process information to develop accurate and stable representations of the current state of the environment (?state representation?). This requires precise neural activity timing synchrony between prefrontal and sensory systems and within prefrontal networks. Our Center focuses on the unifying hypothesis that processes underlying state representation dysfunction are relevant to psychosis, providing a window into pathophysiologic heterogeneity and precision treatment. Four Projects span three species (nonhuman primates, mice, and humans) and eight methodologies (genetic manipulations, slice physiology, ensemble recordings, LFP, behavior, EEG, fMRI, cognitive training). We use a central computational perspective to translate and integrate across species and methodologies: Changes in neural information processing affect parameters underlying attractor dynamics and influence state representation processes. Such changes create observable effects in behavior and neurophysiology, which we can study through the lens of attractor network models to inform our understanding of pathophysiologic heterogeneity, clinical trajectories, and precision treatment. Each Project: 1) Uses the same behavioral tasks to probe components of state representation across species and experiments; 2) Accesses parallel neurophysiologic metrics, with a focus on neural system activity timing, excitatory-inhibitory balance, and noise; 3) Uses advanced data-driven causal discovery analyses to facilitate cross-paradigm integration and novel hypothesis generation. The Projects are supported by a Translational Neurophysiology Core, a Computational Core, and an Administrative Core. Aim 1 investigates behavior and neurophysiology of state representation dysfunctions characteristic of psychosis in a nonhuman primate model of prefrontal network failure in psychosis mediated through NMDA-R signaling (PROJECT 1); in mice with cell type-specific ablation of NMDA-R function and carrying psychosis- associated genetic variants (PROJECT 2); and from an EEG-fMRI study of healthy controls and people with early psychosis (PROJECT 3). Aim 2 develops attractor network models of state representation at multiple levels of detail, incorporating behavioral, synaptic, and cellular microcircuit data from animal neurophysiology studies (PROJECTS 1 & 2) to identify parameters that account for state representation dysfunctions characteristic of psychosis and the behavioral and neurophysiological observations made in humans (PROJECTS 3 & 4). Aims 3 and 4 focus on reliability and predictive significance of state representation dysfunctions in early psychosis, and precision treatment approaches targeting specific dysfunctions.