2019 — 2020 |
Engel, Tatiana |
R01Activity Code Description: To support a discrete, specified, circumscribed project to be performed by the named investigator(s) in an area representing his or her specific interest and competencies. |
Discovering Dynamic Computations From Large-Scale Neural Activity Recordings @ Cold Spring Harbor Laboratory
Project Summary/Abstract How neural activity is coordinated within local microcircuits and across brain regions to drive behavior is a central open question in neuroscience. Recent advances in massively-parallel neural recording tech- nologies are producing dynamic activity maps during complex behaviors, with single-neuron granularity and single-spike resolution. To reveal fundamental dynamic features in these large-scale datasets, new principled and scalable computational methods are urgently needed. To address this need, we will de- velop a broadly applicable, non-parametric inference framework for discovering dynamic computations from large-scale neural activity recordings. Our framework seeks a dynamical model of the data, but unlike existing techniques, does not require a priori model assumptions. Existing techniques commonly ?t data with simple ad hoc models, which often miss or distort de?ning dynamic features. Instead, our non-parametric approach explores the entire space of all possible dynamics in search for the model consistent with the data, and thereby eliminates a priori guess work, ambiguous model comparisons and model-induced biases. We aim to develop optimization algorithms to effectively search through the space of all dynamical models, implement these algorithms on GPUs to achieve maximal computational speed, and derive information-theoretic bounds to quantify reliability of our computational methods. To demonstrate how our novel methods aid scienti?c discovery, we will employ them to examine decision- related activity in parietal and premotor cortices. While different theoretical models of decision-making have been proposed, it still remains unknown how decision computations are implemented on the level of individual neurons and neural populations. Our analyses will offer the ?rst computational models of decision-making rooted directly in neural data, reconcile stability of population dynamics with hetero- geneity of single-neuron responses, reveal differences in decision-computations across cortical layers, and identify differences in decision-related dynamics of excitatory vs. inhibitory neurons.
|
1 |
2021 |
Engel, Tatiana |
RF1Activity Code Description: To support a discrete, specific, circumscribed project to be performed by the named investigator(s) in an area representing specific interest and competencies based on the mission of the agency, using standard peer review criteria. This is the multi-year funded equivalent of the R01 but can be used also for multi-year funding of other research project grants such as R03, R21 as appropriate. |
Multiscale Computational Frameworks For Integrating Large-Scale Cortical Dynamics, Connectivity, and Behavior @ Cold Spring Harbor Laboratory
Project Summary/Abstract A central problem in neuroscience is to understand how activity arises from neural circuits to drive animal behaviors. Solving this problem requires integrating information from multiple experimental modalities and organization levels of the nervous system. While modern neurotechnologies are generating high-resolution maps of the brain-wide neural activity and anatomical connectivity, novel theoretical frameworks are urgently needed to realize the full potential of these datasets. Most state-of-the-art methods for analyzing high-dimensional data are based on detecting correlations in neural activity and do not provide links to the underlying anatomical connectivity and circuit mechanisms. As a result, conclusions derived with these methods rarely generalize across different behaviors and are hard to validate in perturbation experiments. In contrast, mechanistic theories, which combine connectivity, activity, and function, have been highly successful in understanding function of small neural circuits. Conditions under which insights from small circuits scale to large distributed circuits have not been explored. Mechanistic theories informed by multiple data modalities are critically missing to guide experiments probing global neural dynamics on the brain-wide scale. The main goal of this proposal is to develop computational frameworks for modeling global neural dynamics, which utilize anatomical connectivity and predict rich behavioral outputs on single trials. Our project will address two complementary aims. First, we will take advantage of recently available datasets of high-resolution brain- wide neural activity and anatomical connectivity to construct a multiscale model of functional dynamics across the mouse cortex. Integrating measurements across multiple scales, from mesoscopic to near-cellular resolution, we aim to reveal the effective degrees of freedom at each scale, which constrain global neural dynamics and drive rich patterns of behavior. Second, we will leverage techniques from dynamical systems theory and artificial recurrent neural networks to develop circuit reduction methods that infer interpretable low-dimensional circuit mechanisms of cognitive computations from high-dimensional neural activity data. Rather than merely detecting correlations, our method infers the structural connectivity of an equivalent low-dimensional circuit that fits projections of high-dimensional neural activity data and implements the behavioral task. We will apply this method to multi-area neural activity recordings from behaving animals to reveal distributed circuit mechanisms of context-dependent decision making. The computational frameworks developed in this proposal can be validated in perturbation experiments and extended to other nervous systems and behaviors.
|
1 |