Probing neural population dynamics with recurrent neural networks

Large-scale recordings of neural activity are providing new opportunities to study network-level dynamics with unprecedented detail. However, the sheer volume of data and its dynamical complexity are major barriers to uncovering and interpreting these dynamics. I will present latent factor analysis via dynamical systems, a sequential autoencoding approach that enables inference of dynamics from neuronal population spiking activity on single trials and millisecond timescales. I will also discuss recent adaptations of the method to uncover dynamics from neural activity recorded via 2P Calcium imaging. Finally, time permitting, I will mention recent efforts to improve the interpretability of deep-learning based dynamical systems models.

The balance hypothesis for the avian lumbosacral organ and an exploration of its morphological variation

The centrality of population-level factors to network computation is demonstrated by a versatile approach for training spiking networks

Neural activity is often described in terms of population-level factors extracted from the responses of many neurons. Factors provide a lower-dimensional description with the aim of shedding light on network computations. Yet, mechanistically, computations are performed not by continuously valued factors but by interactions among neurons that spike discretely and variably. Models provide a means of bridging these levels of description. We developed a general method for training model networks of spiking neurons by leveraging factors extracted from either data or firing-rate-based networks. In addition to providing a useful model-building framework, this formalism illustrates how reliable and continuously valued factors can arise from seemingly stochastic spiking. Our framework establishes procedures for embedding this property in network models with different levels of realism. The relationship between spikes and factors in such networks provides a foundation for interpreting (and subtly redefining) commonly used quantities such as firing rates.

Dynamic endocrine modulation of the nervous system

Sex hormones are powerful neuromodulators of learning and memory. In rodents and nonhuman primates estrogen and progesterone influence the central nervous system across a range of spatiotemporal scales. Yet, their influence on the structural and functional architecture of the human brain is largely unknown. Here, I highlight findings from a series of dense-sampling neuroimaging studies from my laboratory designed to probe the dynamic interplay between the nervous and endocrine systems. Individuals underwent brain imaging and venipuncture every 12-24 hours for 30 consecutive days. These procedures were carried out under freely cycling conditions and again under a pharmacological regimen that chronically suppresses sex hormone production. First, resting state fMRI evidence suggests that transient increases in estrogen drive robust increases in functional connectivity across the brain. Time-lagged methods from dynamical systems analysis further reveals that these transient changes in estrogen enhance within-network integration (i.e. global efficiency) in several large-scale brain networks, particularly Default Mode and Dorsal Attention Networks. Next, using high-resolution hippocampal subfield imaging, we found that intrinsic hormone fluctuations and exogenous hormone manipulations can rapidly and dynamically shape medial temporal lobe morphology. Together, these findings suggest that neuroendocrine factors influence the brain over short and protracted timescales.

Extracting computational mechanisms from neural data using low-rank RNNs

An influential theory in systems neuroscience suggests that brain function can be understood through low-dimensional dynamics [Vyas et al 2020]. However, a challenge in this framework is that a single computational task may involve a range of dynamic processes. To understand which processes are at play in the brain, it is important to use data on neural activity to constrain models. In this study, we present a method for extracting low-dimensional dynamics from data using low-rank recurrent neural networks (lrRNNs), a highly expressive and understandable type of model [Mastrogiuseppe & Ostojic 2018, Dubreuil, Valente et al. 2022]. We first test our approach using synthetic data created from full-rank RNNs that have been trained on various brain tasks. We find that lrRNNs fitted to neural activity allow us to identify the collective computational processes and make new predictions for inactivations in the original RNNs. We then apply our method to data recorded from the prefrontal cortex of primates during a context-dependent decision-making task. Our approach enables us to assign computational roles to the different latent variables and provides a mechanistic model of the recorded dynamics, which can be used to perform in silico experiments like inactivations and provide testable predictions.

Dynamical System Theory and Mean Field Approximation

Talk & Tutorial

Flexible multitask computation in recurrent networks utilizes shared dynamical motifs

Flexible computation is a hallmark of intelligent behavior. Yet, little is known about how neural networks contextually reconfigure for different computations. Humans are able to perform a new task without extensive training, presumably through the composition of elementary processes that were previously learned. Cognitive scientists have long hypothesized the possibility of a compositional neural code, where complex neural computations are made up of constituent components; however, the neural substrate underlying this structure remains elusive in biological and artificial neural networks. Here we identified an algorithmic neural substrate for compositional computation through the study of multitasking artificial recurrent neural networks. Dynamical systems analyses of networks revealed learned computational strategies that mirrored the modular subtask structure of the task-set used for training. Dynamical motifs such as attractors, decision boundaries and rotations were reused across different task computations. For example, tasks that required memory of a continuous circular variable repurposed the same ring attractor. We show that dynamical motifs are implemented by clusters of units and are reused across different contexts, allowing for flexibility and generalization of previously learned computation. Lesioning these clusters resulted in modular effects on network performance: a lesion that destroyed one dynamical motif only minimally perturbed the structure of other dynamical motifs. Finally, modular dynamical motifs could be reconfigured for fast transfer learning. After slow initial learning of dynamical motifs, a subsequent faster stage of learning reconfigured motifs to perform novel tasks. This work contributes to a more fundamental understanding of compositional computation underlying flexible general intelligence in neural systems. We present a conceptual framework that establishes dynamical motifs as a fundamental unit of computation, intermediate between the neuron and the network. As more whole brain imaging studies record neural activity from multiple specialized systems simultaneously, the framework of dynamical motifs will guide questions about specialization and generalization across brain regions.

Neural Population Dynamics for Skilled Motor Control

The ability to reach, grasp, and manipulate objects is a remarkable expression of motor skill, and the loss of this ability in injury, stroke, or disease can be devastating. These behaviors are controlled by the coordinated activity of tens of millions of neurons distributed across many CNS regions, including the primary motor cortex. While many studies have characterized the activity of single cortical neurons during reaching, the principles governing the dynamics of large, distributed neural populations remain largely unknown. Recent work in primates has suggested that during the execution of reaching, motor cortex may autonomously generate the neural pattern controlling the movement, much like the spinal central pattern generator for locomotion. In this seminar, I will describe recent work that tests this hypothesis using large-scale neural recording, high-resolution behavioral measurements, dynamical systems approaches to data analysis, and optogenetic perturbations in mice. We find, by contrast, that motor cortex requires strong, continuous, and time-varying thalamic input to generate the neural pattern driving reaching. In a second line of work, we demonstrate that the cortico-cerebellar loop is not critical for driving the arm towards the target, but instead fine-tunes movement parameters to enable precise and accurate behavior. Finally, I will describe my future plans to apply these experimental and analytical approaches to the adaptive control of locomotion in complex environments.

Linking dimensionality to computation in neural networks

The link between behavior, learning and the underlying connectome is a fundamental open problem in neuroscience. In my talk I will show how it is possible to develop a theory that bridges across these three levels (animal behavior, learning and network connectivity) based on the geometrical properties of neural activity. The central tool in my approach is the dimensionality of neural activity. I will link animal complex behavior to the geometry of neural representations, specifically their dimensionality; I will then show how learning shapes changes in such geometrical properties and how local connectivity properties can further regulate them. As a result, I will explain how the complexity of neural representations emerges from both behavioral demands (top-down approach) and learning or connectivity features (bottom-up approach). I will build these results regarding neural dynamics and representations starting from the analysis of neural recordings, by means of theoretical and computational tools that blend dynamical systems, artificial intelligence and statistical physics approaches.

Neural manifolds for the stable control of movement

Animals perform learned actions with remarkable consistency for years after acquiring a skill. What is the neural correlate of this stability? We explore this question from the perspective of neural populations. Recent work suggests that the building blocks of neural function may be the activation of population-wide activity patterns: neural modes that capture the dominant co-variation patterns of population activity and define a task specific low dimensional neural manifold. The time-dependent activation of the neural modes results in latent dynamics. We hypothesize that the latent dynamics associated with the consistent execution of a behaviour need to remain stable, and use an alignment method to establish this stability. Once identified, stable latent dynamics allow for the prediction of various behavioural features via fixed decoder models. We conclude that latent cortical dynamics within the task manifold are the fundamental and stable building blocks underlying consistent behaviour.