Henry D I Abarbanel - US grants

9527804 Abarbanel An understanding of the dynamics (or time evolution) of streamflow, its predictability, causality and variability at different time and space scales is a central issue in hydrologic research. The proposed research seeks to develop quantitative notions of predictability and dynamical similarity of streamflow at U.S. sites, using long USGS records (60 to 114 years) of daily streamflow that are presumed to have little human impact. Methods (based on nonlinear dynamics) for forecasting streamflow directly from the time series will be a byproduct of the research. In the context of climate change research, projections of streamflow using exogenous model climate data may be needed for many years in the future. Given uncertain inputs, it is important to assess how quickly, and under which conditions such predictions deteriorate into random traces, irrespective of model formalism. It may also be of interest to know, at least qualitatively, the sensitivity of streamflow to different causative factors. What are the implications for modeling streamflow at different space and time scales? When is it useful to pursue deterministic, distributed or lumped models, and when must one resort to a purely statistical approach? Loosely speaking, streamflow results from the interaction of large scale atmospheric circulation with slowly varying or fixed surface conditions. The latter may lend organization to streamflow as the spatial scale of interest (e.g., drainage area) increases, thus limiting the effective number of "dynamical" factors that have major influence. An idea which we will investigate is that a larger basin spatially averages over the many dynamical processes in the climatic forcing--rainfall at specific locations, geographical features in the basin, evaporation and soil properties which are certainly heterogeneous across any watershed. This averaging may reduce the dimension of the response which we sample by streamflow. Similarly, climatic fluctuations that have "stru cture" at long time scales, (e.g., El Nino Southern Oscillation). may increase stearmflow predictability. An interesting methodology for insights into such processes is provided by recent advances in nonlinear dynamics, and in particular for time series analysis from nonlinear processes. The idea is that time series for a single state variable (e.g., streamflow) for the dynamical system can be used to geometrically reconstruct a "state space" that contains the essential information on predictability and complexity of the underlying system. Predictability is measured in information theoretic terms through Lyapunov exponents that measure the rate of divergence of nearby trajectories in state space. Complexity is measured through generalized dimensions by an examination of how densely the state space is filled and variations in such a density over the state space. Strategies for forecasting the state variable in the reconstructed state space are devised. Our analyses of the Great Salt Lake volume using these methods have been very fruitful, and have given us the firm conviction that significant insights into the nature of the streamflow process and development of theoretical models for streamflow at different scales will be possible. The approach proposed here is to systematically analyze variety of long US streamflow data sets, to (1) see if a reconstruction of the underlying dynamics is possible from the time series, (2) estimate Lyapunov exponents as a measure of predictability, (3) estimate generalized dimensions to describe the complexity of the underlying dynamics, (4) to develop effective forecasting strategies for daily streamflow, and (5) identify how predictability and complexity and forecasting ability vary the climatic attributes and basin attributes such as drainage area. Of particular interest are physical thresholds at which the response of the system undergoes a change to qualitatively different dynamics. The existence of such thresholds is probed after the recovery of sy stem invariants from the time series using nonparametric regression methods with respect to the parameters of interest.

9987614
Terry Sejnowski - University of California, San Diego
IGERT: Graduate Training Program in Computational Neurobiology

This Integrative Graduate Education and Research Training (IGERT) award supports the establishment of a multidisciplinary graduate training program of education and research in computational neurobiology. The goal is to train a new generation of scientists and engineers with a broad range of scientific and technical skills who are equally at home measuring large-scale brain activity, analyzing the data with advanced computational techniques, and developing new models for brain development and function. This integrative training program is centered in the Department of Biology at UCSD and the Salk Institute, but includes faculty members from physics, chemistry, psychology, cognitive science, electrical engineering, computer science, and mathematics, as well as from biology and neuroscience. The training program will give all students hands-on experience in a wide range of advanced experimental and computational techniques through collaborative research between laboratories, industrial internships, and the opportunity to pursue research abroad. The faculty will participate in outreach programs to encourage and prepare underrepresented minorities for a career in computational neurobiology. Research areas in the training program include: (1) synaptic growth and plasticity; (2) neural dynamics; (3) neural population coding; (4) visual perception and memory; (5) stochastic learning algorithms; and (6) functional brain imaging.

IGERT is an NSF-wide program intended to meet the challenges of educating Ph.D. scientists and engineers with the multidisciplinary backgrounds and the technical, professional, and personal skills needed for the career demands of the future. The program is intended to catalyze a cultural change in graduate education by establishing new, innovative models for graduate education and training in a fertile environment for collaborative research that transcends traditional disciplinary boundaries. In the third year of the program, awards are being made to nineteen institutions for programs that collectively span all areas of science and engineering supported by NSF. The intellectual foci of this specific award reside in the Directorates for Biological Sciences; Computer and Information Science and Engineering; Social, Behavioral, and Economic Sciences; Mathematical and Physical Sciences; Engineering; and Education and Human Resources.

This goal of this project is to uncover the fundamental aspects of the biophysics associated with synchronization in small groups of neurons. The question will be addressed both in the biological laboratory as well as using numerical and electronic simulations of the neurons seen in the lab. The research will add to the understanding of how biological nervous systems can process information from the environment and pass it on in a coherent way to decision making centers. The work, while in small nervous systems of invertebrates, will directly bear on how more complex systems can organize to produce interesting functional behaviors. During the course of the project, graduate students and postdoctoral researchers from Physics and Biology will work side-by-side and will learn about the biological sciences as well as the physical processes underlying the biological activity.

This project seeks to uncover the fundamental aspects of the biophysics associated with synchronization in small groups of neurons, in particular to investigate the role of synaptic plasticity, widely believed to underlie memory and learning, in synchronization and information transport in neural circuits. The members of the project will investigate this question both in their biological laboratory as well as using numerical and electronic simulations of the neurons seen in the lab. Graduate students and postdoctoral researchers from Physics and Biology will work alongside the principal researchers and in their training they will learn about the biological sciences as well as the physical processes underlying the biological activity.

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5).

Describing the action of the nervous system in quantitative terms is an important goal in neuroscience, but has been hampered by lack of appropriate techniques to describe the complex time-varying activity patterns of individual neurons. The activity patterns a given neuron expresses depend on the inputs to that neuron and the intrinsic properties of that neuron. A central limitation has been the inability to determine from data collected from real neurons the specific values of various conductances and other properties of those neurons that define their time-varying patterns of activity. To address this limitation, this project will combine real data derived from targeted experiments with modeling of those data to create a new synthesis. Data will be collected under controlled conditions from highly studied identified populations of neurons of the avian forebrain. The experimental procedures especially the structure and families of current pulses in those intracellular recordings will be tightly coordinated with a parallel effort to develop techniques to estimate the parameters of the recorded neurons. The goal of these combined experiments is to fully characterize the parameters of the neurons so as to describe the time-varying properties of the neurons. Success in this effort will be broadly applicable, providing a fundamental new approach to fully incorporating actual experimental data to characterize the activity properties of individual neurons. In the first instance, this will be applied to describing attributes of neuronal activity that are related to vocal production and learning and auditory memory phenomena. This project will also result in significant training opportunities. Biologists early in their careers will be trained in the emerging field of computational neuroscience. Young physicists and other quantitative scientists seeking new problems outside of their traditional disciplines will receive training in biological thinking and biological methods. These individuals will thus learn early in their careers a multidisciplinary approach to scientific problems that is the emerging paradigm for modern research.

In this project the PI will develop a nonlinear dynamical method for the estimation of states and parameters in models of neurobiological systems using experimental data. Utilizing this method the PI will analyze models of individual neurons in the crustacean Pyloric central pattern generator (CPG) and models of sub-circuits of neurons within this CPG. The general problem of determining the parameters of mathematical models of individual neurons and networks of neurons will be addressed using a novel method called dynamical parameter estimation (DPE), which is based on the classical problem of optimal tracking of a desired trajectory. In this case the DPE is known in the control theory literature as a Luenberger observer with adaptive gain: a filter in the sense of Kalman. Using this approach the PI will apply the new technique to the analysis of experimental data from CPG neurons and CPG sub-circuits. Two graduate students will be involved in this efforts. The research group of the PI will also work each summer with Research Experience for Undergraduates (REU) undergraduate students program at UCSD.

Project Summary Changes in synapses and resultant changes in properties of networks of neurons are well established and potent mechanisms for learning. Recent studies have identified an additional potential site of plasticity, regulation of the magnitude of ionic currents in neurons, implicating changes in the intrinsic properties (IP) of the neurons. These will affect the shape of the spike waveform, which is diagnostic, and the trains of spikes a cell will convey to a network, which has functional consequences. The relation between changes in IP and learning remains unclear, however. Here, a recently discovered compelling relation between IP and learning is investigated in the song system of the well-established model birdsong learning. In vitro intracellular recordings of avian forebrain ?HVCx? neurons projecting to the basal ganglia showed that different cells within a given animal shared similarity of waveforms and spike trains emitted in response to current injections, and these differed across animals. Modeling these data in a Hodgkin-Huxley (HH) framework to estimate the magnitudes of five pharmacologically confirmed principal ionic currents revealed that the ion current magnitudes of HVCx from each animal were tightly clustered together but showed large differences across animals. Critically, the differences in HVCX IP between birds was related to the acoustic similarity of their songs, and predictions of this observation were sustained (similarity in sibling animals, developmental changes, inhomogeneity during abnormal singing). Given this suite of unanticipated results, the proposed experiments test the novel hypothesis that sensorimotor feedback errors are transmitted by variability around an IP set point shared across populations of neurons. In the first specific aim, a detailed model relating of HVCx intrinsic properties and features of singing will be developed, relating homogeneity of HVCx IP with song learning by studying juvenile and adult animals singing songs with graded differences (resulting from controlled tutoring during development). The hypothesis that HVCx represent a single neuronal ensemble will be tested by assessing c homogeneity in species singing multiple song types. A second aim will test the hypothesis that sensorimotor feedback errors are transmitted by HVCX variability around an IP ?set point?. Comparing changes in distributions of HVCX IP values when birds change songs in the presence of delayed or pitch shifted feedback will identify if the changes represent song features or errors. Intracellular and multisite extracellular recordings in singing birds or in a fictive singing sleeping preparation will connect in vitro and in vivo properties of the neurons, and determine the time course of changes in IP relative to onset of abnormal singing. A third aim will develop mathematical procedures for estimating all the parameters and the global minima of HH models for each neuron, and a develop a two compartment HH HVCX model including network and IP components, that describes how burst during singing. These experiments aim to identify the cellular and network mechanisms associated with a novel and rapid form of learning mechanism associated with skilled performance.

Project Summary Mechanistically linking network connectivity and the dynamics of neural networks to variation in the behavior of individuals is an overarching goal of neuroscience. Here we address this goal using techniques from network science to calculate functional networks that summarize pair-wise and higher order interactions between all recorded neurons. Network activity will be assessed using sophisticated two-photon (2P) imaging of activity- dependent Ca2+ signaling optimized to maximize the rate of recording and the numbers of neurons recorded. Multineuronal interactions within the networks will be identified, giving rise to encoding models to predict the network activity. Techniques from statistical physics will be used to optimally couple data from intracellular recordings to biologically realistic Hodgkin-Huxley (HH) models representing the contributions of ion currents and other free model parameters of the individual neurons. Networks of HH neurons using model synapses will replace pair-wise correlations to delinate the interrelationships between the ion currents of individual neurons and network interactions and dynamics. Taking advantage of the birdsong learning model, in the proposed experiments these approaches will be applied to the cortical song system HVC nucleus, allowing us to link these scales of investigation directly to behavior. Recent results demonstrate that changes in the intrinsic properties (IP) (ion current magnitudes) of HVC neurons is related to each individual's song, implicating changes within neurons as well as at synapses and networks that are related to learning. Aim 1: fast 2P imaging will be made in brain slices containing HVC that express spontaneous network activity. Model building will be supported by extensive efforts at 3-cell and 4-cell whole cell patch recordings, to better characterize HVC connectivity. The hypothesis that network structure depends on learning will be tested by examining how models vary between individual birds who sang similar or different songs. Models will be extended to in vivo observations by fast 2P imaging in sleeping birds while eliciting fictive singing using song playback, and in singing birds using 1P imaging. Results from the other Aims will further constrain the network and HH model building of Aim 1. Aim 2: the predictive power of the models will be further tested by using cellular resolution 2P optogenetic inhibition of selected neurons in in vivo and in vitro preparations. Aim 3: the role of neuronal IP in shaping network dynamics will be tested by using genetic and viral techniques to transiently modify specific ion channels in specific classes of HVC neurons. Changes in birds' singing behavior will be compared against a predictive HH model relating song structure and ion channel efficacy. Fast 2P imaging in slice and multisite extracellular recordings in singing birds will help to define how IP contribute to network models. Aim 4: single cell gene expression techniques will be used to identify all the HVC cell classes, the ion channels they express, and assess individual variation by examining cohorts of related birds or those singing the same songs. The overall goals and the four Aims are also designed to align with a subsequent U19 application.