Kenneth D. Miller - US grants

Cenozoic climate and ocean chemistry changes are linked to the evolution of deep-water circulation patterns. A project to reconstruct Cenozoic deep-water circulation changes will use benthic foraminiferal . 13C comparisons, supplemented by ongoing faunal (benthic foraminiferal) and sedimentological studies (CaCO3, opal, organic carbon accumulations). An integrated statigraphic framework for synoptic interbasinal and bathymetric (vertical) . 13C comparisons will be constructed. Detailed interbasinal . 13C reconstructions of the Oligocene to Pliocene oceans will delineate various water-type end members (Northern Component, Southern Component, Indian Ocean Water, Pacific Deep Water). Vertical . 13C gradients will be reconstructed in several regions including the North Atlantic and Southern Ocean. These reconstructions will allow evaluation of the role of these critical bottom-water source regions upon the development of long period (106yr) deep-water circulation. The importance of shorter-period (104-105yr) changes across two major climatic transitions (ca. 14-12.5 Ma and 3-2 Ma) will be used to determine if climatic forcing on this bandwidth was linked to deep-water variations as in the Quaternary. %%% This study will involve the use of oxygen and carbon isotopes of foraminifera from sedimentary cores. The history of deep circulation will be inferred for the last 40 million years in order to examine the effects of climate on paleocirculation. The results will provide the opportunity to test models of the interaction of paleoclimate, paleoceanography, and plate tectonic reconstructions on the evolution of the modern ocean.

The upper Oligocene to Miocene section of the U.S. middle Atlantic is ideally suited for the study of changes in relative sea level recorded in passive margin sediments. This award supports a high-resolution multichannel and single channel seismic study of the upper Oligocene to Miocene section along the middle Atlantic continental shelf and slope. These seismic data will be integrated with biostratigraphic and Sr-isotope stratigraphic studies of coastal plain outcrops, slope canyon outcrops, wells, and boreholes. The objective will be to determine the geometry and age of Oligocene to Miocene depositional sequences, and to evaluate the role of relative sea- level changes in developing this record. We will evaluate possible causal links between ice-volume (glacio-eustatic) changes inferred from the deep sea 8180 record and depositional sequences dating from this Oligocene to Miocene "ice house world" will be evaluated. This project will provide the data base needed to plan a continental shelf to slope drilling program that will define precisely the age of these depositional sequences and test models of sedimentation and relative sea-level changes.

This project will develop a chronology for the Paleogene by integrating magnetostratigraphy, stable isotope stratigrajphy, Sr.isotope, and biostratigraphy. the stratigraphic goal is to developa Paleogene time scale, and to test biostratigraphic correlations between low and high latitudes. In addition, the work has the paleoceanographic goal of reconstructing deep.and surface.water fluctuations by makeing synoptic faunal and isotope comparisons across two climat transitions, the Paleocene/Eocene boundary and the late middle Eocene to early Oligocene.

The goals of this proposal are to contribute to the understanding of the circuitry and synaptic mechanisms underlying the function of the cerebral cortex. Since this is the brain structure responsible for much of our conscious perception and action and of our "higher" mental functions, the implications for mental health of such understanding are profound. Excitatory transmission in the cerebral cortex is mediated primarily by glutamate receptors. The N-methyl-D-aspartate (NMDA) receptor is one of at least two subtypes of these. It is unique among glutamate receptors in that the channel it activates has a voltage-dependent conductance under physiological conditions, and the currents it activates have a very slow time course. Recent work has revealed that excitatory transmission in visual cortex depends critically on NMDA receptors. This proposal's aim is to theoretically examine the effects of this dependence on the computational properties of individual visual cortical cells and on model visual cortical networks. This will be accomplished through biophysically detailed computer modeling. Specifically, the aims are: (1) To develop a limited set of candidate biophysical models for NMDA-receptor mediated synaptic activation, through detailed single-cell modeling, constrained by observations at both the biophysical and the physiological level; (2) To combine these NMDA models with existing models of visual cortical circuitry, in order to develop unified, consistent, testable models capable of explaining experimental observations of visual cortical spatio/temporal response properties.

DESCRIPTION (adapted from applicant's abstract): The long-term goal of this work is to understand the circuitry of the cerebral cortex and the rules underlying its activity-dependent development. Primary visual cortex (V1) of the cat is studied as a model system for understanding cortex more generally. Computational modeling is used to determine what patterns of circuitry can account for the functional response properties of V1 neurons and what rules of activity-instructed synaptic modification can yield the self-organization of these patterns of circuitry. Cerebral cortical circuitry underlies most sensory perception, much of motor planning, and most of the higher cognitive functions associated with human intelligence, so an understanding of cortical circuitry and its development will strongly impact our understanding of both normal and diseased brain function. In particular, understanding of V1 circuitry and development will impact our understanding of normal vision and of central diseases of vision such as amblyopia and strabismus. The specific aims of this work are to develop biologically identifiable and testable models of the circuitry of layer 4, the input-recipient layer, of cat V1 and of the development of that circuitry. Studies of development will test the hypothesis that spike-timing-dependent plasticity (STDP), based on spontaneous patterns of activity that exist before visual experience impacts development, can account for the organization of V1 receptive fields and functional circuits. A particular focus will be to understand the development of direction selectivity and of the associated cortical circuitry. Studies of the mature circuit will build on previous work showing that a "correlation-based" circuit, in which excitatory cells tend to project to cells with similar or well correlated receptive fields (overlapping ON- and OFF-subregions) and inhibitory cells tend to project to cells with roughly opposite or anticorrelated or antiphase receptive fields, can account for many of the functional response properties of V1 layer 4 cells. This work will be extended to incorporate new experimental findings on the roles of voltage noise in V1 responses, of orientation-untuned complex inhibitory neurons, and of synaptic depression in V1 responses. It will also be extended to address direction selectivity by incorporating diversity of temporal response properties of input neurons and by extending the spatial correlation-based circuitry to circuitry based on spatiotemporal correlations.

There are a number of reasons why it is desirable to study the simultaneous responses of many cells at a single site in the mammalian neocortex. Such studies can determine the relative response fields and connectivity of nearby cells, yielding insight into local circuits and the origin of cortical response properties. Such studies can also test the idea that neural information may be encoded in synchronization or correlation of cell responses. Such insights into cortical function are fundamental to our understanding of normal mental function and its pathologies. The purpose of the present proposal is to develop accurate, automatic, quantifiable, objective, and real-time means of discriminating many cells in extracellular recording at a single site. This will be done by combining two methods, each of which has recently achieved significant progress on this discrimination problem: the tetrode method of recording, and Bayesian statistical methods of inferring the distinct waveforms underlying a recording. The tetrode method provides additional information, relative to single-electrode recording, which simplifies discrimination; while the Bayesian methods provide optimal and quantifiable discrimination based on the full information available in the recording. The combination of the two methods will produce a method that can discriminate 5-10 cells at a single site in a quantifiable, reproducible, and largely automatic manner, opening to a new quantity and quality of extracellular recording in the cerebral cortex.

The basis of this project is to improve the computing facilities available for undergraduate Math majors at East Central University. East Central University, located in rural Oklahoma, has a strong Teacher Certification program and is also ranks highly in terms of the number of Bachelor degrees granted to native Americans. The Department of Mathematics currently has very limited and antiquated computing facilities, which denies students the opportunity of an education which exposes them to tech- nology, such as that which they will subsequently encounter in industry or graduate school. This lack of exposure to computing technology also impedes students' competitiveness in seeking employment upon graduation. With updated computing equipment, Mathematics majors, and especially those in the Honours program, will be able to incorporate modern technology into their studies, thereby enriching their educational experience and better preparing them for future encounters with technological innovation.

Description (provided by applicant): We seek to understand the circuitry of visual cortex, the rules underlying its activity-dependent development, and the computational functions these subserve. These understandings are critical to our understanding both of normal vision and of its central disorders including amblyopia and strabismus. We focus on V1 as a model system. We use modeling to test coherent hypotheses as to the circuitry underlying V1 functional response properties and the plasticity rules underlying circuit development. Modeling allows testing of an integrated picture of circuit structure and/or developmental rules against a variety of experimental results and the development of new and unanticipated tests of that picture. The specific aims of this project are: (1) To create biologically identifiable and testable models of the mature circuitry of visual cortex. In particular, we will test the hypotheses (i) That intracellular as well as extracellular classical RF responses in LGN-recipient V1 simple cells, including a contrast-dependent decrease in voltage noise that is critical to contrast-invariant orientation tuning, can be quantitatively understood from simple, essentially feedforward models. (ii) That the apparent determination of the response tuning properties of LGN-recipient V1 simple cells by their feedforward input and the strong recurrence seen in V1 can be integrated into a coherent circuit model of layers 2/3 and 4 under the hypothesis, for which we provide evidence, that the recurrence functions as an inhibition-stabilized network: a network in which excitatory recurrence alone is strong enough to cause instability, but the circuit is stabilized by feedback inhibition. We will determine the conditions under which this circuit can account for classical and extra-classical receptive field properties and the structure observed in spontaneous activity, clearly isolate the contribution of the recurrence to responses, and develop novel tests of this architecture. (2) To create biologically identifiable and testable models of the development of the circuitry of visual cortex. In particular, studying the critical period (CP) for monocular deprivation in mouse V1, we will develop a unified model of homeostatic and Hebbian CP plasticity, and theoretically test the hypothesis that the induction of the CP by maturation of inhibition occurs due to an increase in the ratio of visual activity to spontaneous activity caused by that maturation. PUBLIC HEALTH RELEVANCE The visual cortex is the brain structure with which we see, creating our visual perception from the information provided by the eyes. To understand both how we see normally and how central visual disorders such as strabismus and amblyopia arise and can be treated, it is critical to understand how the circuitry of visual cortex processes visual information and is organized by visual experience. These are our research aims.

This project studies the effects of incorporating, into deep neural networks for visual processing, several heretofore unincorporated features of biological visual cortical circuits. Deep neural networks are artificial circuits loosely inspired by the brain's cerebral cortex. Their abilities to solve complex problems, such as recognizing objects in visual scenes, have revolutionized artificial intelligence and machine learning in recent years. The hierarchy of layers in a deep network trained for visual object recognition also provides the best existing models of the hierarchy of areas in the visual cortex implicated in object recognition (the "ventral stream"). This project seeks to understand whether and how incorporating additional features of brain circuits may (1) improve machine learning performance, particularly on tasks that are more challenging than those typically studied; and (2) yield improved models of visual cortex. Improving the performance of deep networks would yield great benefits across wide swaths of society and industry that are impacted by advances in artificial intelligence. Improved models of visual cortex will advance understanding of cortical function, which may lead to significant further benefits for understanding normal mental functioning and perception and their potential enhancement, as well as mental illness and perceptual and cognitive deficits.

Deep networks currently achieve their success using almost purely feedforward processing. Yet the visual cortical ventral stream that helped inspire deep networks also uses massive recurrent processing within each area as well as feedback connections from higher areas to lower areas and "bypass" connections from lower areas to areas multiple steps higher in the hierarchy. Deep networks also use "neurons" that can either excite or inhibit different neurons that they project to, whereas biological neurons are exclusively excitatory or inhibitory. This project will incorporate feedback and bypass connections into deep networks, as well as local recurrent processing in networks of separate excitatory and inhibitory neurons. Recent work by the investigators has shown how local recurrent processing explains a number of nonlinear visual cortical operations often summarized as "normalization." Simple forms of normalization currently used in deep networks maintain activities in an appropriate dynamic range, but the biological forms of normalization involve interactions between different stimulus features and locations in determining neural responses, which may have important computational roles e.g. in parsing visual scenes. The performance of deep networks incorporating these features will be assayed on a variety of visual tasks and as models of ventral stream neural data and human psychophysical data, and compared to performance of existing deep net models.

Understanding how a healthy brain interprets sensory signals and guides actions, and why an unhealthy brain fails to perform these functions properly, is a profound and ambitious goal of 21st century science. Integrating knowledge of neural circuit function into a coherent picture of perception, cognition and action requires extraordinary cooperation and coordination between three research areas: experimentation, data analysis and modeling. The National Science Foundation Theory Team at Columbia University will unite exceptional resources in statistical data analysis and theoretical modeling with an extensive network of experimental collaborators to address the enormous challenges facing neuroscience. Never has the need been greater for theoretical insights and sophisticated data analysis. The field of neuroscience is facing a torrent of complex data from a system that is, itself, extraordinarily complex. Future progress requires developing the ability to extract knowledge and understanding from these data through analyses and modeling that capture the essence of what they mean. The goal of the NeuroNex Theory Team at Columbia is to establish, through the quality of its research, the excellence of its trainees, and the impact of its visitor, dissemination, and outreach programs, a new cooperative paradigm that will move neuroscience to unprecedented levels of discovery and understanding.

High-density electrode recording, wide-field calcium imaging and complex connectivity mapping are bringing neuroscience into an era of extensive multi-area and even whole-brain studies of neural activity and circuitry. The neuroscience community desperately needs new ways of interpreting data obtained from different species using myriad techniques and for thinking about neural processing over large length and time scales and across multiple brain areas. In response to these challenges, two major goals will drive and define research at the NeuroNex Theory Team at Columbia: first, integrating the analysis methods and theoretical models used to infer meaning from data with each other and with the experiments that generate these data; and second, providing analytic tools and theoretical frameworks to understand interactions between multiple brain regions and to draw important overarching lessons from experiments exploiting a variety of techniques across different species. Progress will be made through a tight integration of theoretical techniques with outstanding experimental collaborators working on a variety of systems and species. Graduate and postdoctoral training will stress technical excellence and broad perspectives in both theoretical and experimental neuroscience. Outreach will be made to other researchers through visitor and exchange programs, sponsored meetings and dissemination of research results and high-quality, user-friendly software. Outreach will be made to the broader community by sharing the excitement of neuroscience research with elementary and high school students and with the general public. This NeuroNex Theory Team award is co-funded by the Division of Emerging Frontiers within the Directorate for Biological Sciences, the Division of Physics and the Division of Mathematics within the Directorate of Mathematical and Physical Sciences, and by the Division of Brain and Cognitive Sciences within the Directorate of Social, Behavioral and Economic Sciences, as part of the BRAIN Initiative and NSF's Understanding the Brain activities.

Summary A fundamental problem of neuroscience is understanding the operation of cerebral cortical circuits. Given the basic similarity of all cortical circuitry despite many differences across species and areas, understand- ing of any particular cortical circuit will be a major step toward that goal. Here we propose to bring an extremely strong team of theorists together to model the circuitry of mouse primary visual cortex (V1) in unparalleled depth, in tight interaction with experimentalists who will produce transformative data to inform and test our models. We will initially focus on understanding contextual modulation and its modulation by running and arousal in layer 2/3 processing, incorporating the three best-studied subtypes of inhibitory neurons, parvalbumin- (PV), somatostatin- (SOM), or vasoactive-intestinal-peptide-expressing (VIP) interneurons, and possible subtypes of SOM neurons. We will also develop tractable single-compartment models of dendritic inhibition, which will be a critical advance allowing network models to address the function of different interneuron types targeting different neuronal compartments while remaining simple enough to yield insight. We will study the impacts on network behavior of SOM inhibition at dendrites vs. PV inhibition on soma and of the short-term plasticity of synapses in the system. We will then advance to incorporating further subtypes, addressing a wider range of dynamic response properties, and modeling layer 4 and the full system of layers 2 through 4, building on the extensive data gathered by experimental projects in this proposal. Finally, working with Project 1, we will develop a uni?ed model of mean stimulus responses and correlated ?uctuations, and address V1 responses to natural stimuli. To understand the functions of cortical specializations such as cell subtypes and layers, we must not only systematically incorporate structure revealed in the data, but use modeling approaches aimed at gaining insight, e.g. understanding mechanisms that produce speci?c activities, or the forms of circuit modulation that can result from targeting particular cell types in particular combinations. To achieve this, we will gradually, step-by-step, add complexity to our models, understanding at each step what new behaviors are introduced, what greater structure or alterations occur in previously understood mechanisms, and what new mechanisms become visible. The most innovative aspect of this proposal is that we will use theoretical approaches designed to give in- sight into mechanisms to grapple with the complex speci?c details of mouse V1. Existing approaches typically either study more abstract models (e.g., generic excitatory and inhibitory cells) or put all known details (along with, necessarily, a great many unknown ones) into the computer with the belief that this will reproduce brain activity, an approach unlikely to generate functional responses or testable predictions. Our approach promises to dramatically deepen our insight into the mechanisms of processing in cortex and in mouse V1 in particular.

Abstract - Administrative Core (Columbia University) The Administrative Core will be a key facilitator in integrating the multi-institutional, interdisciplinary V1 team of theorists and experimentalists that includes researchers from five institutions: Columbia University, California Institute of Technology, University of California, Berkley, University of California, San Francisco and the University of Pittsburgh. To meet this goal, the Administrative Core will coordinate and integrate all U19 Program functions; guide and facilitate interactions between the Project and Core leaders, Principal Investigators, and research staff; ensure that the U19 V1 Team maximizes the utilization of existing and established resources; facilitate data sharing and collaboration with key NIH BRAIN Initiative stakeholders and the greater neuroscience research community; and provide rigorous and regular fiscal oversight of the research projects and cores. The Administrative Core leaders will convene members of the Internal Advisory Committee and External Scientific Advisory Board, who are experienced research scientists with proven track records of leadership, to provide expert oversight of the U19 Program?s infrastructure, scientific direction, and resource development for the research community. The Administrative Core will also develop a website to use as a tool for communication and collaboration with our program components and the larger research community.

A fundamental goal for understanding the brain and mammalian and human intelligence, and to understand how processing goes awry in genetic and developmental diseases, is to understand the principles of operation of cerebral cortex. A key step is to understand canonical operations carried out by cortex. Here we will explore the operations of cortical circuitry in experiments guided by a new theory of a candidate canonical circuit operation. Sensory cortex must globally integrate localized sensory input to parse objects and support perception. In individual neurons, this manifests as modulation of responses to local stimuli by context or top-down influences such as attention and as interactions between local stimuli in driving responses (normalization). These interactions tend to be suppressive for stronger stimuli but more weakly suppressive or facilitative for weaker stimuli. Recent theoretical work in Dr. Miller's lab has proposed a novel cortical circuit motif, the stabilized supralinear network (SSN), that provides a simple unified explanation for a wide variety of neural responses related to global integration. The model serves as a guide for new experimental explorations of cortical circuitry in Dr. Van Hooser's laboratory, using both traditional experimental recording techniques and his recently developed novel optical methods for manipulating cortical activity with high spatial and temporal resolution. The SSN model, if successful, will be elaborated to best explain experimental results. In Aim 1, the light-activated channel channelrhodopsin2 (ChR2) and an optical stimulation system are used to drive activity of cortical circuits in precise spatial and temporal patterns to test the contribution of cortical circuits to normalization and contextual modulation including various SSN predictions about them. In Aim 2, the balance of drive to excitatory (E) vs. inhibitory (I) cells within the cortex will be altered using viruses that largely restrict expression of ChR2 to E or I cells. This will test SSN model predictions involving modulation of network gain by modulatory input biased toward E or I cells, mechanisms of attentional modulation, and the dependence of a paradoxical result -­ adding drive to I cells reduces steady-state I responses -- on the spatial pattern of drive to I cells and level of cortical activation. RELEVANCE (See instructions): We will test the predictions of a powerful framework for understanding how sensory cortex globally integrates multiple sources of input, bottom-up and top-down, to produce neuronal responses and ultimately perception. Understanding circuit changes that cause breakdown of this cortical operation may provide insight into disorders such as autism and schizophrenia, which show deficits in contextual or global processing. Understanding global integration will be necessary for the creation of prosthetic devices to treat blindness and other disorders.

Understanding the cerebral cortex requires data-based theoretical models that can yield in- sight into the circuit mechanisms of cortical computation, and reproduce detailed cortical dynamics across stimuli and brain states. The primary visual cortex (V1) is the best-studied cortical area by both theorists and experimen- talists, yet current models - whether statistical or circuit based ? only poorly capture how V1 neurons respond to complex stimuli, such as natural scenes. The ultimate goal of this team project is to obtain the necessary experimental data and build the detailed circuit-based models that explain how V1 circuits encode natural visual stimuli. In so doing, we aim not only to provide a mechanistic understanding for how V1 dynamics forms the basis of vision, but also to establish a more generalizable paradigm for understanding any cortical area. Our assumption is that current models fall short for two reasons: on the experimental side, we are still missing most of the fundamental details about the synaptic connectivity and physiological responses of V1 cell types; while on the theory side, prevailing circuit-based models reduce V1 to just a few cell types, and either capture the static responses of V1 neurons to simple stimuli but not their trial to trial ?uctuations, or capture ?uctuations, but not their rich array of non-linear responses properties that are central to visual computation. Our hypothesis is that we can achieve a circuit-based model that explains cortical responses and dynamics to natural stimuli by implementing the following three steps: 1) identify and incorporate all the differentiable V1 neuronal cell types into our model; 2) measure and incorporate the synaptic connectivity and intrinsic properties of these cell types; 3) measure and accurately predict the visual responses of each of these cell types to diverse visual stimuli and in multiple brain states. We focus on circuit-based rather than statistical models of V1 for two reasons: they can provide insight into neural mechanisms of visual computation and the regimes of cortical operation, and because they will permit us to test their accuracy by validating their predictions for how V1 responds to de?ned experimen- tal perturbations. To implement these perturbations, we will employ multiphoton holographic optogenetics, which allows us to manipulate V1 circuits with the level of precision formerly only possible in the realm of theory. Here we bring together an outstanding team of theorists, experimentalists, and data scientists to leverage cutting edge new brain mapping technologies that we will use to build and validate dramatically improved models of visual cortical function and dynamics.

Project Summary  Despite the enormous complexity of the brain, it is becoming increasingly apparent that structures like the  cerebral cortex are modular, relying on a set of canonical computations that occur across brain regions and  modalities to mediate perception, cognition and behavior.  One important example of a canonical computation  is the summation of various driving, contextual, and modulatory neuronal inputs to yield spiking output. The  question of how cortical networks integrate these inputs and transform them into spiking outputs of individual  neurons is of central importance to neuroscience.  A significant challenge to understanding these computations  is that each neuron is embedded within a larger circuit of neurons, each modulating one another?s activity.  So,  understanding how a particular neuron responds to input necessarily involves understanding the larger circuit.   Recent optogenetic studies have found different patterns of input summation in mouse vs. monkey V1.  Recently developed theoretical models have produced specific predictions about the differences in network  circuitry that can lead to differences in summation, and predict how summation non-­linearities depend on  inputs to the network. The proposed research will test these predictions and seek to understand these circuit  computations using a combination of theoretical work and optogenetic modulation of circuits in mouse and  monkey.  Aim 1: Varying E and I optogenetic stimulation and visual contrast independently to measure  spike response summation to multiple inputs. In this Aim, theoretical models of input summation across  varying cortical circuit regimes will be developed, and recently developed optogenetic tools will be used in  awake mouse and monkey V1 to test predictions generated by these models and identify the corresponding  regimes. The optogenetic tools include a new viral strategy that directs expression of different opsins to  inhibitory vs. excitatory neocortical neurons in the macaque. Simultaneous and independent activation of E and  I and the visual stimulus, all within this theoretical framework, will enable us to test whether observed  differences in summation properties reflect fundamental species differences or reflect a common computation  operating in different parameter regimes.  Aim 2: Determine the circuit elements controlling dynamics of  cortical network responses using dynamic optogenetic stimulation. In this Aim, experiments using  dynamic optogenetic and visual stimulation patterns and theoretical analysis of the models with dynamic inputs  will be used to elucidate the temporal dynamics of summation. Aim 3: Determine if different inhibitory  subclasses control different aspects of input integration. Different inhibitory subclasses will be stimulated  optogenetically to decipher their respective roles in input summation. Taken together, these Aims will help  define the roles played by excitatory and inhibitory neurons in mediating summation of neuronal inputs to yield  spiking output. This information will be critical for understanding brain disorders associated with failures in  perception and attention, as is seen with autism, schizophrenia, and Alzheimer?s disease.